Messenger No142

The Messenger
No. 142 December 2010

I
m
p
r
o
v
i
n
g

n
e
t
w
o
r
k

l
i
n
k
s

t
o

P
a
r
a
n
a
l
T
h
e

V
L
T

A
d
a
p
t
i
v
e

O
p
t
i
c
s

F
a
c
i
l
i
t
y
M
u
l
t
i
p
l
e

p
o
p
u
l
a
t
i
o
n
s

i
n

g
l
o
b
u
l
a
r

c
l
u
s
t
e
r
s
D
i
s
s
e
c
t
i
n
g

t
h
e

s
u
p
e
r

c
l
u
s
t
e
r

W
e
s
t
e
r
l
u
n
d

1
2
The Messenger 142 December 2010
The communications technology
To guarantee current needs while making
provision for future performance demands,
the EVALSO project has opted for optical
fbre network paths that could be procured
from commercial installations or, where
none existed, created new ones. The use
of dark fbres (i.e. physical fbres whose full
capacity is available) and, where this is not
economically possible, use of reserved
wavelengths, provides an optimal and fex-
ible infrastructure for the traffc manage-
ment needs of today, with a clear potential
for future growth. In order to exploit these
optical communications most effectively,
DWDM (dense wavelength division multi-
plexing) technology was selected for the
communications equipment.
The optical path infrastructure (numbers
refer to Figure 1) consists of:
new fbre cables (1 and 4) that serve the
Paranal Observatory and OCA sites up
to Ruta 5 at La Varilla;
a pair of fbres from the existing installa-
tion along Ruta 5 to relay from La Varilla
to Antofagasta;
fxed wavelength communication (2)
between the telecommunications pro-
vider Point of Presence (TELCO PoP) in
Antofagasta and Santiago;
dark fbres (5) between the TELCO PoP
and the end points in Santiago, namely
ESOs Vitacura offces and the REUNA
offce in Providencia;
and housing space for EVALSO equip-
ment at the TELCO PoP in Antofagasta
(6) and Santiago (3).
In order to connect the two observatories
to the Chilean TELCO infrastructure,
new cables had to be installed in the Ata-
cama Desert for a total of about 100 km.
The cables were of the type used for
direct underground installation and the frst
cable, approximately 80 km long, was
laid down between Paranal and the point
of connection to the telecommunications
providers backbone, situated along the
Panamericana (Ruta 5, the Pan-American
Highway) at La Varilla. The second, ap -
proximately 20 km long, is between OCA
and an intermediate point on the frst
cable. These installations established an
optical fbre path between the two obser-
vatories and, together with the portion
of fbre procured from the commercial net-
work, to Antofagasta.
Giorgio Filippi
1
1
ESO
EVALSO (Enabling Virtual Access to
Latin-American Southern Observatories)
is an international consortium of nine
astronomical organisations and research
network operators, part-funded under
the European Commission FP7, to create
and exploit high-speed bandwidth con-
nections to South American observato-
ries. A brief description of the project
is presented. The EVALSO Consortium
inaugurated a fbre link between the
Paranal Observatory and international
networks on 4 November 2010 capable
of 10 Gigabit per second.
It should come as no surprise that the
remoteness which makes a site excellent
for opticalnear-infrared astronomy also
conficts with the need for good con-
nections to a fast communications infra-
structure. With the constant increase
in data rates and, more generally, in com-
munications needs, limits in bandwidth
may impede the effciency of operations,
and hinder future expansion. For the
ESO Paranal Observatory and the Ruhr-
Universitt Bochum Observatorio de Cerro
Armazones (OCA), this confict will soon
cease to exist. On 4 November 2010, at
the ESO Vitacura offces in Santiago, the
EVALSO Consortium formally inaugurated
a fbre-based system capable of 10 Giga-
bit per second (Gbps) to connect the two
observatories to the international aca-
demic networks.
The partners of EVALSO are the Universit
degli Studi di Trieste, Italy; the European
Southern Observatory; Ruhr-Universitt
Bochum (RUB), Germany; Consortium
Gestione Ampliamento Rete Ricerca
(GARR), Italy; Universiteit Leiden, the
Netherlands; Istituto Nazionale di Astro-
fsica (INAF), Italy; Queen Mary, University
of London, United Kingdom; Cooperacin
Latino-Americana de Redes Avanzadas
(CLARA), Uruguay; and Red Universitaria
Nacional (REUNA), Chile. More details
of the overall project and its members are
available at the EVALSO website
1
.
The project focuses on two aspects:
building an infrastructure to connect
observatory sites effciently to the
European astronomical community by
linking to the network infrastructures
created in recent years with EC support
(in particular ALICE
2
and GEANT
3
);
promoting research activities to enable
and validate new ways to interact
with the remote facilities made available
by high-bandwidth communications.
More details can be found in Filippi et al.
(2010).
The present article focuses on the com-
munications infrastructure, which is
broken down into two major components:
the optical paths and the communica-
tions equipment. The locations served by
the new high capacity infrastructure are:
the ESO Paranal Observatory and OCA,
about 1200 km north of Santiago; the
REUNA offces in Antofagasta (about
120 km from Paranal); and the ESO and
REUNA offces in Santiago de Chile.
The Organisation
Enabling Virtual Access to Latin-American
Southern Observatories
Atacama
Dark bre: unlimited capacity
Lambda: (currently) 10 Gbps
1
2
3
4
5
6
7
Figure 1. Overview of the EVALSO communications
infrastructure. See text for details.
3
Underground installation is not only more
secure, but limits the visual impact on
the pristine desert environment. A special
machine cuts a trench about 20 cm wide
and 80 to 120 cm deep, depending on the
nature of the soil (see Figure 2). The cable
is suitable for direct installation without
ducts (Figure 2). As reels of 4 km length
are used, splicing boxes have to be placed
in chambers at the same distance. The
construction also crosses existing roads
and a gas pipeline. A movie of the cable-
laying process is included in the ESO pod-
cast
4
. The distance from the end of the
new cable, at the La Varilla crossroads,
to the nearest installation of the telecom-
munications provider in Antofagasta is
about 50 km. This distance has been cov-
ered by procuring a fbre pair from the
existing facility. A fxed wavelength channel
(LAMBDA), capable of carrying 10 Gbps,
links Antofagasta and Santiago.
EVALSO equipment is installed at fve
nodes. At Paranal this node receives the
traffc to and from both observatories.
For each user community separate sub-
channels are used to transport the traffc
to its fnal destination. The aggregation
point for the traffc handled by REUNA and
from the observatories is sited at Antofa-
gasta. Here the bundled (but not mixed!)
traffc is passed to the provider equipment
for the approximately 1200 km section
over the LAMBDA cable to Santiago. The
LAMBDA termination in Santiago is the
distribution point for the traffc coming
from the northern region of Chile. Using
local dark fbres, the Paranal traffc is
delivered to the ESO Vitacura offces and
the REUNA traffc to the REUNA offces
in Santiago, and from there to the com-
mercial and academic networks.
Project history and inauguration
The EVALSO project team (SA1 work
package) began work in 2008 by taking
over an initial market survey, which gave
enough confdence in terms of technical
and economic feasibility to continue with
the project. Based on this input, the SA1
team detailed the specifcation of the fnal
system and entered the procurement
phase, assigning the optical paths to ESO
(in coordination with OCA for their part
of the connectivity) and the DWDM equip-
ment to REUNA. The complex procure-
ment phase took nearly the whole of 2009,
leading to the fnal decision and the start
of engineering in the last quarter. The
installation of the new cables as well as
the confguration and integration of the
existing commercial parts, both paths and
DWDM equipment, kept the team busy
until October 2010, when the frst trans-
mission test could be made.
The EVALSO formal inauguration took
place in a ceremony at the ESO Offces
in Santiago (see the picture on the Astro-
nomical News section page 40). The event
was attended by a number of ambassa-
dors and diplomats from EU countries, a
very large delegation from the EC, high
level offcials from ESO, ALMA and other
organisations. The event was opened by a
short speech by the ESO Director General,
Tim de Zeeuw, and there were speeches
from Jos Palacios (REUNA president), Rolf
Chini (RUB-OCA), Massimo Tarenghi (ESO
representative in Chile), Mario Campolargo
(EC) and Ambassador Fernando Schmidt
Arizta (vice-minister of Foreign Affairs
of Chile). Fernando Liello, Universit degli
Studi di Trieste, EVALSO Project Coordi-
nator, explained the EVALSO framework
and its relations with the overall aca -
demic research networks. The technical
aspects of the EVALSO infrastructure were
illustrated by Giorgio Filippi, ESO, for the
optical paths, and Sandra Jaque, REUNA,
for the DWDM equipment.
The overall system is now available for
the research activities of the EVALSO
Consortium. There are three key areas that
the project will develop:
Fast and effcient access to data: at pre-
sent, data collected at the Paranal Ob-
servatory can take anywhere from hours
to weeks before it becomes available for
use in Europe. EVALSO will allow data to
be transferred in near real time.
Virtual presence at the observatories:
the bandwidth offered by EVALSO
makes a virtual presence at the obser-
vatories possible.
New possibilities in observing: EVALSO
can serve several of the needs of its
partner members, for example, by oper-
ating robotic telescopes on Cerro
Armazones, supporting their use for
educational purposes and improving the
scientifc exploitation of the capabilities
of both observatories through innova-
tive operations schemes, which will act
as pathfnders for the operation of future
facilities like the European Extremely
Large Telescope.
As a member of the Consortium, ESO
is planning to integrate the system into its
communications infrastructure in the
coming months, leading to its use in regu-
lar operation of the telescopes on Paranal.
Results of these developments will be
reported in a future article.
References
Filippi, G. et al. 2010, Proc. SPIE, 7740, 77401G
Links
1
EVALSO web page: http://www.evalso.eu
2
ALICE network: http://www.redclara.net
3
GEANT pan-European network: http://www.geant.
net/About_GEANT/pages/home.aspx
4
ESO Podcast: http://www.eso.org/public/videos/
eso1043a/
Figure 2. Two views of the work involved in laying the
EVALSO optical cable from Paranal to link with the
existing network infrastructure; upper image: cutting
the trench; lower: laying the cable in the trench.
T
e
l
e
s
c
o
p
e
s

a
n
d

I
n
s
t
r
u
m
e
n
t
a
t
i
o
n

The Adaptive Optics Facility deformable sec-
ondary mirror, a Zerodur 1.1-metre lightweighted
mirror, shown in testing. See Arsenault et al.
(p. 12) for details.
Eight ALMA antennas observing as an inter-
ferometer at the high site on the Chajnantor
plateau. See Testi et al. (p. 17) for more details. C
r
e
d
i
t
:

A
L
M
A

(
E
S
O
/
N
A
O
J
/
N
R
A
O
)
5
Gunther Witzel
1
Andreas Eckart
1
Rainer Lenzen
2
Christian Straubmeier
1
1
I. Physikalisches Institut, Universitt
Kln, Germany
2
Max-Planck-Institut fr Astronomie,
Heidelberg, Germany
As expected for a Nasmyth instrument,
NAOSCONICA shows a signifcant
instrumental polarisation that is strongly
dependent on the parallactic angle
of the source and can reach up to 4%
in the degree of linear polarisation.
Detailed modelling of the polarimetric
properties of the optical components
of NAOSCONICA using the Stokes/
Mueller formalism allows the instrumen-
tal polarisation to be corrected with
an accuracy of better than 1% in linear
polarisation. In addition we propose an
observation strategy that is expected to
allow instrumental polarisation effects
to be corrected to an accuracy of about
0.1%.
Polarimetric measurements can provide
important information on the nature of
radiative processes, which completes the
picture obtained from total intensity ob-
servations. In particular this is the case
for the supermassive black hole at the
centre of the Milky Way and its associ-
ated variable near-infrared source Sagit-
tarius A* (Sgr A*). Here, as for other tar-
gets, the interpretation of the polarimetric
data depends crucially on the quality
of the polarisation calibration and the
knowledge of instrumental systematic
errors. The majority of polarimetric obser-
vations of Sgr A* have been conducted
with the near-infrared imager NAOS
CONICA (NACO) at the VLT, and we re-
port here the results of a comprehensive
analysis of the instrumental polarisation
(IP) of this instrument. Full numerical
details can be found in Witzel et al. (2010).
NACO and its polarimetric mode
NACO is the adaptive optics near-infrared
imager at VLT Unit Telescope 4 (UT4,
Yepun), consisting of the adaptive optics
module NAOS and the camera CONICA.
This instrument provides a mode for
polarimetric differential imaging combin-
ing a Wollaston prism and a /2 wave
plate (a half-wave plate, HWP). The Wol-
laston prism separates the incoming par-
tially polarised light into two orthogonal,
linearly polarised outgoing beams (the
ordinary and extra-ordinary beams). The
HWP enables the observer to change
the angle at which the measurement is
conducted without rotating the Wollaston
prism with respect to the detector and
thus without changing the feld of view.
Intensity measurements over at least four
angles (0, 45, 90 and 135) in two inte-
grations (two orthogonal beams simulta-
neously) are necessary to obtain infor-
mation about linear polarisation. Circular
polarisation cannot be measured, since
the polarimetric mode of NACO does not
include a /4 wave plate.
After about eight years of successful
operation of NACO and plenty of polari-
metric observations of the Galactic Cen-
tre (GC), Sgr A*, and of many other tar-
gets, it now seemed plausible to analyse
the polarimetric mode on the basis of
public ESO archive data. Long polarimet-
ric light curves of bright sources at the
GC have been especially useful in investi-
gating the dependency of the IP on the
telescope position, and in determining
the accuracy achieved to date and attain-
able for future observations.
A model of the instrumental polarisation
To understand the IP of NACO we have
developed a model of the instrumental
effects within the Stokes/Mueller formal-
ism. The polarisation of the incoming light
is mainly affected by refections at metal-
lic coatings of mirrors within the light path
of NACO. Each refection causes cross-
talk between the four Stokes parameters
that describe the polarisation state. For
unpolarised incoming light this crosstalk
generates, for example, linear polarisa -
tion orthogonal to the plane of incidence.
The magnitude of the crosstalk depends
strongly on the angle of incidence: only
mirrors with an inclination of > 20 will
contribute signifcantly to the IP. All polari-
metric effects of the optical elements and
their relative orientation can be described
by linear operations on the Stokes vector,
the so-called Mueller matrices.
As a Nasmyth focus instrument, NACO
shows an instrumental polarisation
that depends strongly on the parallactic
angle of the observed source. The cause
of the variable part of the IP is the fold-
ing mirror M3 that is tilted at 45 (see Fig-
ure 1). While NACO itself is de-rotated
with respect to the source, the orientation
of M3 and, therefore, also the orientation
of its IP with respect to the polarisation
of the observed source, changes with the
parallactic angle. As shown in Figure 1,
other mirrors within NAOS and CONICA
Telescopes and Instrumentation
On the Instrumental Polarisation of NAOSCONICA
M2
M3
NAOS
/2 p|ate
Wo||aston
Detector
S13/27 opt|cs
M1
Figure 1. Optical elements of VLT UT4, NAOS and
CONICA and their relative orientation are sketched
at the moment of the meridian transit.
6
to three, mainly describing the variable
part of the IP that is known with an accu-
racy of about 0.1% (see Figure 3). A dis-
advantage is the alteration of the feld of
view that results from a rotation of NACO
as a whole as compared to the full feld.
Polarised emission of Sgr A*
Figure 4 shows the derived total and
polarised fux, linear polarisation and
polarisation position angle for Sgr A*
measured with NACO in June 2006. As a
result of our analysis we can show that
the common boot-strapping calibration of
time series of Sgr A* on the basis of the
average foreground polarisation as deter-
mined by Knacke & Capps (1977) is in
good agreement with our new calibration.
The main reason is that the statistical
errors of the photometry dominate the
described instrumental effects for short
integrations of faint sources like Sgr A*.
Only at the lowest states of polarisation
do both methods deviate signifcantly
in polarisation angle, an effect that can
contribute as well. Since these mirrors
are part of the de-rotated instrument their
contribution is, however, constant.
An important effect is caused by the
HWP. A maintenance inspection revealed
that the actual reference system for
the HWP position angle is turned by 6.6
+/ 0.2 with respect to the reference
assumed in the NACO manual. This re-
sults in an angle offset of 13.2 for
measurements of the polarisation angle
that has to be compensated for.
A simulation of the polarisation degree of
the IP as a function of hour angle (for a
source at the position of Sgr A*) is shown
in Figure 2. The maximum instrumental
polarisation is about 4% (all effects, in-
cluding the optical elements of CONICA)
and drops signifcantly at the time of me-
ridian transit. A comparison of our model
with the observed time series of the
Stokes parameters of bright sources at
the GC and with standard star observa-
tions confrms the model (see Figure 3).
The polarisation of standard stars can
be calibrated with an accuracy of better
than ~0.5% for the Q and U Stokes
parameters.
To our knowledge our results for the
IRS16 stars at the GC are the frst polari-
metric measurements of sources within
the central parsec of the GC, since
Knacke & Capps (1977), that are inde-
pendently calibrated using a method that
goes beyond boot-strapping proce-
dures. An example is shown in Figure 3.
The accuracy for sources of this bright-
ness (m
k
> 9.5 mag) is better than ~1%
in linear polarisation and better than
~5 in polarisation angle for polarisation
degrees higher than ~4%.
A strategy for high precision polarimetry
The remaining systematic uncertainties of
about 1% in linear polarisation result from
insuffcient knowledge of the characteris-
tics of the Wollaston prism (such as the
relative transmission of the orthogonal
channels) and the effects of the fat-feld
calibration (the calibration light for the fat
feld is potentially polarised intrinsically
up to 1%). A common calibration method
to circumvent problems like this is chan-
nel-switching. In this method, at four
additional angles with a 90-offset with
respect to the above-mentioned four
angles intensities are measured. With
this additional information Stokes param-
eters that are corrected for parts of the
IP can be calculated. In particular, the
effects of the Wollaston and the fat-feld-
ing can be compensated for by this strat-
egy.
Unfortunately this calibration is not suita-
ble to correct all instrumental effects. In
particular, the effects of refections at
metallic surfaces cannot be eliminated
completely and reference offsets as
caused by the HWP remain uncompen-
sated. But we can show that this method
is very useful if: 1) the HWP offset is al-
ready compensated for during observa-
tion; 2) the HWP is used to switch be-
tween the 0/90 and 45/135 angle
pairs (respectively between their corre-
sponding orthogonal pairs); and 3) NACO
as a whole is rotated to switch by 90 for
the measurement of the additional four
angles. With this strategy the number of
parameters of our model can be reduced
Witzel G. et al., On the Instrumental Polarisation of NAOSCONICA
-2 -1 0
0
1
1
P
o
|
a
r
|
s
a
t
|
o
n

d
e
g
r
e
e

(
%
}
Hour ang|e (ha}
2
2
3
3
4
4
-2 -1 0
-5
1

(
%
}
Hour ang|e (ha}
2
0
3 4
5
Figure 2. The polarisation degree of
the instrumental polarisation as pre-
dicted by the model is shown as a
function of hour angle. Linear polarisa-
tion (in green) and total polarisation
(including the circular polarisation, in
red) are shown for an unpolarised
source with (dashed line) and without
(solid line) the systematic effects of the
analyser and fat-felding.
Figure 3. Stokes Q (in blue) and U (in
green) parameters as a function of
hour angle for IRS16C (linear polarisa-
tion 4.6% at PA18, m
k
= 9.55). The
solid line is the best
2
-ft of the model
and the points represent observa-
tions from 2009. The ftting confrms
the time dependency of the instru-
mental polarisation as predicted by
the model and additionally allows
the apparent polarisation of this star to
be determined.
7
be explained by the interplay of the HWP
offset and the boot-strapping method
that cannot correct for it. In these polari-
sation states a reliable estimation of the
polarisation angle is impossible anyway,
and we can conclude that for Sgr A* in its
bright fare phases the boot-strapping
calibration yields the same results within
the statistical uncertainties as a calibra-
tion with our more elaborate polarisation
model. Only these bright phases have
been interpreted in the framework of
the relativistic modelling (Zamaninasab et
al., 2010; Eckart et al., 2006; Meyer
et al., 2006) that highlights the infuence
of strong gravity in the vicinity of the
4 10
6
M
A
supermassive black hole at
the centre of the Milky Way.
Acknowledgements
We gratefully acknowledge fruitful discussions
with Christian Hummel and the ESO User Support
Department.
References
Eckart, A. et al. 2006, A&A, 455, 1
Knacke, R. F. & Capps, R. W. 1977, ApJ, 216, 271
Meyer, L. et al. 2006a, A&A, 460, 15
Meyer, L. et al. 2006b, A&A, 458, L25
Witzel, G. et al. 2010, A&A, in press
Zamaninasab, M. et al. 2010, A&A, 510, A3+
Figure 4. Light curve of Sgr A* observed with NACO
in June 2006. Upper panel: total fux (black) and
polarised fux (blue); middle panel: polarisation posi-
tion angle; lower panel: linear polarisation. For the
polarimetric parameters, the results of both calibra-
tion methods are shown: boot-strapping (red points)
and the model described here (black points).
0
0
10
P
o
|
.

d
e
g
r
e
e

(
%
}
20
30
40
50
60
70
50 100 150
T|me (m|n}
200
0 50 100 150 200
0 50 100 150 200
-90
-60
-30
P
o
|
.

a
n
g
|
e

(
d
e
g
}
0
30
60
90
120
0
5
10
F
|
u
x

(
m
J
y
}
1 June 2006
15
20
Image of NACO at the Nasmyth focus of VLT UT4
(Yepun). NACO consists of the adaptive optics
module NAOS mated with the near-infrared imager
and spectrograph CONICA.
8
Following tests on the instrument in
November 2009 and February 2010,
VIMOS was removed from the telescope
between May and July 2010, so that
the frst stage of the upgrades could be
made. The instrument was re-commis-
sioned at the end of July. Making fne
adjustments and optimising the system,
then took a few more weeks.
New shutters and other measures to
improve reliability
The exposure is controlled by 100 mm
iris shutters manufactured by the Prontor
company. Due to the size and speed
requirements, these shutters are subject
to considerable wear that has led to
occasional failures. A failure occurring at
the end of science integration is espe-
cially frustrating because it means that
the exposure is lost and must be counted
as technical downtime. Analysis of night
reports from the last two years shows
that the technical downtime due to shut-
ter or detector controller failures has
been 0.9% on average, a non-negligible
fraction of the total VIMOS downtime
which varied between 6% (Q4/2009) and
10% (Q4/2008).
Since this type of shutter is no longer
commercially available, ESO has reverse-
engineered the Prontor shutters and
built ten copies. A new shutter controller
was also designed, which has no dissipa-
tion in the open or closed condition and
has an electromagnetic braking function
to reduce the mechanical loads. Proto-
types have survived lifetime tests of up to
200 000 operations without degradation.
Four of these new shutters have been
mounted and no technical downtime due
to shutter failures has been experienced
since then.
The Mask Exchange Unit (MEU) of each
of the four VIMOS arms consists of
four mechanisms (mask selector/gripper/
translator/blocker) that must be precisely
aligned for reliable operation. About
half of the total technical downtime of
VIMOS is due to MEU failures. In the frst
quarter of 2010 we began performing
daytime dry runs with the MEU whenever
new masks were mounted. The idea
was to detect possible failures during
daytime, when they can still be corrected.
Peter Hammersley
1
Lise Christensen
2
Hans Dekker
1
Carlo Izzo
1
Fernando Selman
1
Paul Bristow
1
Pierre Bourget
1
Roberto Castillo
1
Mark Downing
1
Nicolas Haddad
1
Michael Hilker
1
Jean-Louis Lizon
1
Christian Lucuix
1
Vincenzo Mainieri
1
Steffen Mieske
1
Claudio Reinero
1
Marina Rejkuba
1
Chester Rojas
1
Rubn Snchez-Janssen
1
Alain Smette
1
Josefna Urrutia Del Rio
1
Javier Valenzuela
1
Burkhard Wolff
1
1
ESO
2
Excellence Cluster Universe, Technische
Universitt Mnchen, Germany
The high multiplex of the VLT visible
imager and multi-object/integral-
feld spectrometer, VIMOS, makes it a
powerful instrument for large-scale
spectroscopic surveys of faint sources.
Following community input and recom-
mendations by ESOs Science and
Technology Committee, it was decided
to upgrade the instrument in phases.
The frst phase of the upgrade is
described and included changing the
shutters, installing an active fexure
compensation system, replacing the
detectors with CCDs with a far better
red sensitivity and less fringing, and
improving the data reduction pipeline.
VIMOS (Figure 1) is a powerful visible
(360 nm to 1000 nm) imager and
mult-object/integral-feld spectrometer
mounted on VLT Unit Telescope 3
(Melipal). Following a workshop on spec-
troscopic surveys
1
and recommenda-
tions by the ESO Science and Technol-
ogy Committee, it was decided that
the instrument, which entered operation
in 2002, should be upgraded. In order
to miminise downtime, it was decided to
complete the upgrade in phases. This
article describes the activities of the frst
upgrade and the resulting improvement in
performance.
The instrument and upgrades
VIMOS has four identical arms, each with
a 7 8 arcminute feld of view on the sky
with a gap between the felds of 2 arc-
minutes. The instrument offers three main
observing modes:
U-, V-, B-, R-, I- and z-band imaging
covering four felds each 7 8 arc-
minutes in size.
Slit-based multi-object spectroscopy
with spectral resolutions from a few
hundred to 2500 in each of the four
imaging felds.
Integral-feld unit (IFU) spectroscopy
with felds of view between 13 13 and
54 54 arcseconds.
After eight years of operations, and to
extend its useful life, it became necessary
to upgrade the instrument in order to
address various issues. The frst upgrade
included:
1) replacing the shutters, which were
worn out, and improving the reliability
of the instrument;
2) replacing the CCD detectors;
3) reducing the instrument fexure;
4) improving the data reduction pipeline.
Upgrading VIMOS
Figure 1. VIMOS is shown on the Nasmyth platform
of VLT UT3.
9
Apart from the extra manpower to per-
form the daytime tests, the experience
with dry runs is encouraging, although
more statistics are needed.
The new detectors
The original VIMOS detectors were
thinned back-illuminated CCDs made by
e2v. These were cosmetically very clean
and gave a good performance in the
blue and visible wavelength ranges. At
wavelengths longer than about 700 nm,
however, the detectors had strong fring-
ing, which meant that near 850 nm the
quantum effciency (QE) of the detector
could change by up to 40% for changes
in wavlength of about ten nanometres,
or a movement across the detector of a
few pixels. This made obtaining high
quality spectroscopy very diffcult in this
wavelength range, particularly when there
was fexure present.
The new detectors are also e2v devices.
These are the same format as the origi -
nal but the silicon is more than twice
as thick. This has dramatically reduced
the fringing so that now the maximum
change in QE is at most 2% and cannot
be seen at all with the LR-Red grating
or in imaging. Figure 2 shows two raw
stellar spectra near 850 nm taken with
the old (red) and new (black) detectors
using the HR-Red grating. With the old
detector the fringing is so large that none
of the stellar features can be rec ognised,
and so careful reduction is required to
produce usable results (Scodeggio et al.,
2009). The raw spectrum taken with the
new detector, however, is almost good
enough to use, so simplifying the reduc-
tion and improving the quality of the re-
sults, especially on faint red objects.
Figure 3 shows the change in imaging
zero points between the old and new
detectors. The signifcant improvement in
the R-, I- and z-bands can be clearly
seen, however there is a slight decrease
in sensitivity in the U-band. The data
used for calculating the zero point
with the new detectors were taken just
before the primary mirror of Melipal
was recoated, which has improved the
zero points by 0.05 to 0.1 magnitudes.
The active fexure compensation
VIMOS is a large instrument, weighing
three tonnes, and suffers from 1220 pix-
els of fexure between the focal plane and
the detector, when the instrument is ro-
tated to follow the feld rotation. This fex-
ure causes the image to smear during
long exposures, makes calibration more
diffcult, and reduces the accuracy of
object position measurements taken in
the imaging mode, necessary for MOS
mask production. The instrument was
delivered with a passive (mechanical)
fexure compensation system. Its perfor-
mance was improved by ESO to about
four pixels, but the system is not easy to
maintain or adjust, and in 2009 some
arms exhibited fexures of nearly six pix-
els when the instrument rotated. These
fexures had a major impact on opera-
tions. The spectroscopic fat and arc cali-
brations had to be taken immediately
after the science observations to ensure
that the fexures affecting science and
calibrations were as similar as possible.
Furthermore, the relative pointing of the
four arms between pre-image and spec-
troscopic observation could change,
thus offsetting the sources in the slit. This
was particularly annoying, as observers
could never optimally position the targets
in all four quadrants at the same time.
Therefore, it was decided to install an
active fexure compensation system
(AFC). Two motors were placed on the
fold mirror of each arm, allowing the
0
850 860 870 880
10
4
2 10
4
C
o
u
n
t
s
3 10
4
4 10
4
5 10
4
Wavelength (nm)
Figure 2. Raw stellar spectra taken
with the old (red) and new (black)
detectors using the HRred grism. The
star observed with the old detector
was over a magnitude brighter that the
star observed with the new detector.
The noise in the red spectrum is
caused by the fringing which makes
the detector response vary rapidly
with wavelength. This fringing has al-
most disappeared with the new de -
tector so that the absorption features,
such as the CaII triplet lines, can be
easily seen.
0
0.5
0.5
1
Filter
U
1
2
3
4
B V R I z
C
h
a
n
g
e

i
n

z
e
r
o

p
o
i
n
t

(
m
a
g
.
)
Figure 3. The change
in imaging zero point (in
magnitudes) between
the old and new detec-
tors is shown. The
results for each of the
four arms (indicated 1-4)
are shown by separate
points.
10
exposures. Using the new recipes is
mandatory for reducing data obtained
after the VIMOS CCD upgrade, and they
can also be used for reducing older data.
This new software, developed at ESO,
is intended to replace the original set of
fve MOS recipes.
The new calibration recipe uses a calibra-
tion approach, based on pattern rec-
ognition, which was already applied suc-
cessfully to the FORS1/2 and EFOSC2
pipelines (Izzo et al., 2007). This greatly
reduces the software maintenance work-
load as it no longer requires any prelimi-
nary optical and spectral modeling of
the instrument. This approach has been
adopted in order to cope with the me-
chanical instabilities affecting any real-
world instrument, a problem which was
especially felt with VIMOS. Supporting
the new VIMOS mosaic commission-
ing would have been impossible using
the old MOS pipeline. VIMOS, with its
four quadrants and six grisms, required
the manual recomputation of at least
72spectral distortion models at each
major instrument intervention. With the
new recipes, this is no longer necessary.
The new recipes also signifcantly im -
prove the accuracy of the wavelength cal-
ibration, the quality of the sky sub traction,
image of the focal plane to be displaced
in X and Y on the detector. A fbre in the
focal plane, but just outside the nominal
imaging feld, then acts as a reference
source and so before an observation is
started the image of the focal plane can
be correctly positioned on the detector.
Any fexure during an observation is
corrected by driving the motors using a
look-up table. With the current system
the target positioning can be done with
approximately onepixel accuracy, while
the registration between science and
calibration observations is approximately
two pixels, the latter degraded by hystere-
sis. Duringthe commissioning it was grat-
ifying to be able to see all targets centred
in all quadrants at the same time!
The new MOS pipeline
In the new release of the pipeline data
reduction software (version 2.5.0, deliv-
ered to the users in October 2010) two
new MOS recipes have been added; one
for processing the calibration frames,
and another for reducing the scientifc
Hammersley P. et al., Upgrading VIMOS
Figure 4. A close-up of
the motors of the active
compensation system.
8.5
8.8
9.1
9.4
1
2

+

l
o
g
(
O
/
H
)
27
Figure 5. The image
shows the galaxy
NGC6754 observed
with the HST/NICMOS
in the J-band. From the
two IFU pointings the
oxygen abundance was
determined in the
regions corresponding
to where four SNe
exploded (shown as
open circles).
11
the optimal extraction of the detected
objects, and the sky fringing correction.
The new pipeline and the manual can be
downloaded
2
.
The commissioning
The commissioning of an instrument as
complex as VIMOS needs to be carefully
planned. There are many modes, and the
tree of dependencies has become even
more complex with the introduction of the
AFC system. The AFC reference source
is at the edge of the feld and so focusing
is now even more critical if defocus is
not to cause the centroids to move. The
conversion from pixel coordinates to milli-
metres at the mask plane, is now tied
to the coordinates of the AFC reference
pixel. Any change to the reference pixel
means that the mask-to-CCD matrix
must be re-determined.
VIMOS has three reference systems
which must be aligned to a few pixels in
4000; the detector xy in pixels, the focal
plane XY in mm and the spectral dis-
persion direction. This last alignment is
particularly important for multiplexed ob -
servations with the low resolution grisms,
as the reduction pipelines assume that
the contaminating zero or second orders
are well aligned with the frst order spec-
trum (we thank Luigi Guzzo, Bianca
Garilli, and Marco Scodeggio for strongly
insisting on this point). There are four
arms, and six grisms per arm, each of
which needs to be aligned, and so realign-
ment of all grisms is a lengthy process.
The re-characterisation of the instrument
after any intervention that causes a dis-
placement of the image of the focal plane
on the detector, takes a minimum of a
week. During the commissioning, which
was also a time to become familiar with
the newly installed hardware, we had
to re peat this re-characterisation a few
times. Although this led to a loss of
effciency during the frst two months of
service observing, the stability achieved
in the coeffcients of the mask-to-CCD
matrix are now outstanding, with daily
positioning variations well below a pixel.
Performance validation
As part of the commissioning, perfor-
mance validation observations were
made to help characterise the MOS and
IFU modes and provide a taste of the
performance after the upgrade. Figures5
and 6 show the results from the IFU
observations. Figure 5 shows the HST/
NICMOS J-band image of NGC6754
where four supernovae were observed
between 1998 and 2005. Two belonged
to massive stars that exploded in core
collapse events (TypeII SNe) and the
other two to white dwarf star explosions
(Type I a SNe). NGC 6754 is dominated by
numerous HII regions and widespread
star formation over the face of the galaxy.
The strong optical emission lines in the
spectra at the two VIMOSIFU pointings
allows the oxygen abundance in the re-
gions where the SNe exploded to be
derived, as well as in other regions of the
galaxy. The oxygen abundance was cal-
culated from the H/[NII] ratio using the
calibration of Pettini & Pagel (2004). The
colour maps on the right-hand side of
Figure5 illustrate this abundance, where
the dark blue colour corresponds to solar
metallicities, green and red to super-solar
metallicities, and violet and black to sub-
solar metallicities. Figure 6 shows an
extracted spectrum from a single spaxel
identifying the lines used, but also dem-
onstrating fewer systematic sky subtrac-
tion residuals beyond 700 nm, thanks to
the absence of detector fringes. The sig-
nal-to-noise ratio for continuum emis-
sion is 50% higher, while in the region of
strong sky emission lines the improve-
ment is a factor of two, relative to the old
detectors.
Next steps
The frst stage of the upgrade has sig-
nifcantly improved the performance
of VIMOS in the red and reduced the
amount of fexure to values close to the
initial specifcations of the instrument.
Even so, further improvements are being
planned for 2011 and 2012:
1. Replace the high resolution blue grism
by a volume phase holographic grating
with almost double the effciency.
2. Improve the reliability and accuracy
of the grism placement, mask insertion
and focus mechanisms.
3. Increase the speed of the grism
exchange unit (expected increase in
overall operational effciency is 2%).
4. Eliminate the need to perform pre-
imaging with VIMOS before masks can
be made. Potentially this could in -
crease operational effciency by 10 to
15%.
After these changes we expect to have an
instrument with a much improved stability,
reliability, and effciency that will permit
more ambitious science to be attempted.
References
Izzo, C., Jung, Y. & Ballester, P. 2007, in The 2007
ESO Instrument Calibration Workshop, 191
Pettini, M. & Pagel, B. E. J. 2004, MNRAS, 348, 59
Scodeggio, M. et al. 2009, Messenger, 135, 13
Links
1
Conference web page including presentations:
www.eso.org/sci/meetings/ssw2009/
2
VIMOS reduction pipeline available at:
http://www.eso.org/pipelines
Figure 6. The VIMOS
spectrum of one single
spaxel of the IFU obser-
vation of NGC6754.
5000 6000 7000
H
[NII] 6583
H
[OIII] 5007
R
e
l
a
t
i
v
e

u
x
Wavelength ()
8000 9000
0
1
2
3
4
5
6
7
12
improve the ensquared energy in the
K-band by a factor of two over the
7.5-arcminute feld of view of HAWK-I.
The four LGS WFSs are mounted on
a rotating bearing inside the Hawk-I hous-
ing, while a natural guide star tilt sensor
picks up a star in a ring outside the
Hawk-I feld of view. An additional natural
guide star WFS is used for calibration
and maintenance operations of the
deformable secondary mirror (DSM).
The GALACSI and GRAAL real-time
computers share identical SPARTA hard-
ware architecture and most software
features. SPARTA is a standard platform
for real-time applications developed by
ESOs Adaptive Optics Department and
uses a hybrid architecture of a feld pro-
grammable gate array, digital signal pro-
cessor and central processing unit. It is
designed to be fexible by offering sophis-
ticated software control and post-pro-
cessing features while accommodating
various combinations (numbers) of WFSs
and deformable mirrors. All WFS cam-
eras and tilt sensor cameras are identical
and use as detectors e2v CCD220
devices controlled by the New General
Controller (NGC; Baade et al., 2009)
developed by the ESO Optical Detector
Team.
Status of the project
GALACSI
The GALACSI AO module design and
manufacture is led by the ESO AO
Department with the support of other
ESO divisions (the Technology Division,
Integration Department of the Instru-
mentation Division and Software Devel-
opment Division). The main mechanical
structure and optics have been out-
sourced to industry and off-the-shelf
components are being purchased for the
electro-mechanisms. The control soft-
ware is being developed in-house and
integration will be undertaken by the Inte-
gration Department. The mechanisms
are controlled via the remote motor con-
troller switching system based on the
controller area network protocol. The fnal
design review was held in June 2009
and several important procurements were
launched at that stage. GALACSIs main
structure, manufactured by Bossenkool
(the Netherlands), has gone through fnal
Robin Arsenault
1
Pierre-Yves Madec
1
Jrme Paufque
1
Stefan Strbele
1

Jean-Francois Pirard
1

lise Vernet
1

Wolfgang Hackenberg
1

Norbert Hubin
1

Liselotte Jochum
1

Harald Kuntschner
1

Andreas Glindemann
1

Paola Amico
1

Miska Lelouarn
1
Johann Kolb
1
Sbastien Tordo
1

Robert Donaldson
1

Christian Snke
1

Domenico Bonaccini Calia
1

Ralf Conzelmann
1

Bernard Delabre
1

Mario Kiekebusch
1

Philippe Duhoux
1
Ivan Guidolin
1
Marco Quattri
1
Ronald Guzman
1
Bernard Buzzoni
1
Mauro Comin
1
Christophe Dupuy
1
Jutta Quentin
1
Jean-Louis Lizon
1

Armin Silber
1

Paul Jolly
1

Antonio Manescau
1
Peter Hammersley
1

Javier Reyes
1

Andreas Jost
1
Michel Duchateau
1

Volker Heinz
1

Clmentine Bechet
1

Remko Stuik
2
1
ESO
2
Huygens Laboratory, University of
Leiden, the Netherlands
The Very Large Telescope (VLT) Adap-
tive Optics Facility is a project that
will transform one of the VLTs Unit Tel-
escopes into an adaptive telescope
that includes a deformable mirror in its
optical train. For this purpose the sec-
ondary mirror is to be replaced by a thin
shell deformable mirror; it will be pos-
sible to launch four laser guide stars
from the centrepiece and two adaptive
optics modules are being developed
to feed the instruments HAWK-I and
MUSE. These modules implement inno-
vative correction modes for seeing
improvement through ground layer
adaptive optics and, for high Strehl ratio
performance, laser tomography adap-
tive correction. The performance of
these modes will be tested in Europe
with a custom test bench called ASSIST.
The project has completed its fnal
design phase and concluded an intense
phase of procurement; the year 2011
will see the beginning of assembly, inte-
gration and tests.
Overview
The ESO VLT Adaptive Optics Facility
(AOF) is upgrading one of the Unit Tele-
scopes (UTs) with a new secondary
mirror (M2) unit. This unit includes an
adaptive secondary mirror (with 1170 ac -
tuators), four laser guide stars (LGS)
formed by four 22 W sodium beacons
launched from the telescope centrepiece
and a dedicated adaptive optics (AO)
instrument park to provide users with
optimised AO correction modes. The two
AO components are GALACSI, the AO
module for MUSE that will provide ground
layer adaptive optics (GLAO) correction
and a laser tomography correction for
MUSEs narrow-feld mode for high Strehl
ratio correction in the visible spectral
range, and GRAAL, the AO module for
HAWK-I, which will provide only ground
layer adaptive optics correction. The
Adaptive Secondary Simulator and
Instrument Testbed (ASSIST) will be used
for a complete test phase of the AOF in
Europe. This project constitutes a major
stepping stone towards the European
Extremely Large Telescope (E-ELT).
GALACSI is a module mounted at the
Nasmyth focus; it contains four LGS
wavefront sensors (WFS), a natural star
tilt sensor and an infrared low-order sen-
sor. The laser guide stars and WFSs can
be tuned to two different felds of view: a
1-arcminute feld of view for GLAO cor-
rection, producing a gain of two in ens-
quared energy at 750 nm; and a 7.5-arc-
second feld of view for laser tomography
correction, providing a high Strehl ratio in
the visible (5% at 650 nm) on-axis.
GRAAL is similar in design to GALACSI,
but only applies a GLAO correction to
Progress on the VLT Adaptive Optics Facility
13
acceptance and was delivered to ESO in
October 2010; it is shown in Figure 1.
The optics is being manufactured by
SESO (France) and delivery is expected
in December 2010. Several smaller
mechanical components have already
been received and are being tested and
some are shown in Figure 2. The feld
selector is a critical key system and the
positioning stages manufactured by
Physik Instrumente have been received
and are currently being tested. The elec-
tronic control is progressing steadily
and several VME racks are complete and
have been integrated into the cabinets.
The assembly and integration of GALACSI
started in the ESO integration hall dur-
ing the autumn of 2010 and will continue
throughout 2011. The stand-alone test
for the GALACSI module (without the
DSM) will begin in 2011 to validate the
whole assembly without the other major
subsystems of the AOF. The complete
system test phase for GALACSI with the
DSM on ASSIST is expected to start in
early in 2013 after GRAAL has completed
its corresponding phase.
GRAAL
The GRAAL module is also led by the AO
Department with support from the same
departments as GALACSI, but with
one main difference: a major contract has
been issued for the important task of
manufacturing and integrating the main
structure and opto-mechanics, including
all stages (provided by ESO) and the
WFS co-rotator. This contract is being
undertaken by NTE SENER S.A. (Spain)
and will be completed in early 2011. This
scheme will give the integration activities
a strong start in Garching. The optics
is also being manufactured by SESO and
delivery is expected in December 2010.
Several software modules are being
developed in common for GRAAL and
GALACSI and there are synergies be -
tween both. During summer 2010, a VLT
standard software module was deliv-
ered to NTE SENER S.A. with the corre-
sponding hardware to control the co-
rotator (consisting of an internal bearing
and drive system to rotate the four LGS
WFSs) and is being used by NTE SENER
S.A. for their integration and testing pro-
cess, as shown in Figure 3.
When the GRAAL main structure is deliv-
ered to Garching early in 2011, the optics
will be integrated with the rest of the
electronics and software. GRAAL tests in
stand-alone mode should start by late
2011 to prepare for a system test phase
with the DSM on ASSIST in 2012.
Deformable secondary mirror
The deformable secondary mirror is being
outsourced to MicroGate and ADS
(Italy) and is the heart of the Adaptive
Optics Facility. This contract completed
its fnal design review in December
2007 and assembly started at the end of
2009. Several key components have
been received by the suppliers and work
is progressing along two parallel paths
at MicroGate, where the electronics and
software are being developed, and at
Figure 1. GALACSIs main mechanical structure
is shown during acceptance at Bossenkool
(the Netherlands) in September 2010.
NFM Dichr. unit Beam Splitter unit
Comm. Camera
Field Selector
LGS WFS
assembly
Calibration unit
CU mirror
To MUSE
To vis TT sensor
from VLT
To WFS
Figure 2. Several
smaller mechanical
components of
GALACSIs optical
train are shown.
Figure 3. GRAALs main structure is shown being
integrated and tested at NTE SENER S.A. in Spain.
14
consist of 1178 nm infrared fbre Raman
lasers, doubled in frequency by second
harmonic generation to 589 nm (see
Bonaccini Calia et al. [2010] for details);
Figure 7 shows the test laser setups.
These lasers have very few components,
require minimal optical alignment and
occupy a small volume, all contributing
to a greater reliability. This represents
a major milestone in terms of simplicity,
compactness and improved reliability.
The current LGS system uses the
PARSEC laser, which delivers 5 W of Na
power and sits on a large optical bench
located in the laser cleanroom below
the Nasmyth platform of UT4. In contrast
the TOPTICA 22W laser fts in a volume
of half a cubic metre (excluding the elec-
tronics cabinets)!
TNO Science & Industry (the Nether-
lands) holds the second important con-
tract. They will fnalise the reference
design provided by ESO and manufac-
ture the optical tube assembly, which
is the telescope part of the launch tele-
scope system (see Figure 8). This is
based on a two-lens design, the largest
being 40 cm in diameter while a 30 cm
beam is launched into the atmosphere.
A 45-degree mirror steers the beam and
allows the laser guide star to be pointed
to the appropriate off-axis distance.
The ESO 4LGSF team is focusing on
the rest of the laser system design: beam
diagnostics and jitter control for fast
beam pointing; optical and mechanical
design; electronic and software design;
and analysis of the whole system. Inter-
facing the 4LGSF with the UT and the
ADS where the electro-mechanics is
being assembled and manufactured. The
main highlight of 2010 has been the
delivery of the reference body, a Zerodur
1.1-metre lightweighted optical compo-
nent that provides the reference surface
for the thin shell mirror, shown in Figure 4
(and featured in an ESO announcement
1
).
SAGEM (France) is continuing its efforts
on the thin shell contracted to them
by MicroGate. The convex side has been
polished successfully to specifcations
and the thinning process from the back-
side is starting. Note that SAGEM is
proceeding carefully and made a detailed
analysis of the thinning process before
fnalising the details. Figure 5 shows the
integration and service stand for the
DSM and M2 hub. Figure 6 shows some
of the electronic control boards for the
DSM. The next step is a milestone in mid-
2011 to review the progress of the inte-
gration. Delivery of the DSM to Garching
is planned for early 2012, when the
system test phase in Europe will begin.
The Four Laser Guide Star Facility
The Four Laser Guide Star Facility (4LGSF)
project includes two important con-
tracts contributing to the design and
manufacturing of this system. TOPTICA
Photonics, a company located near
Munich, is responsible for the laser de -
sign and manufacture. TOPTICA has
been selected as the preferred supplier
after an initial call for tender and a com-
petitive preliminary design with another
supplier. The manufacturing contract
was signed in June 2010. A fnal design
review will be conducted in June 2011
and six months later a pre-production
unit will be ready so that all the laser
specifcations can be verifed. Then
the four units will be produced. They
Arsenault R. et al., Progress on the VLT Adaptive Optics Facility
Figure 4. Part of the
DSM reference body
made by SESO in
France (see the image
on p. 4, upper, for the
whole DSM). The intri-
cate rib structure is
literally sculpted from a
monolithic block of
Zerodur. The 1170 circu-
lar holes allow passage
for the actuators. This
process, known as light-
weighting, reduces the
mass by 80% to a mere
35 kg.
Figure 5. Integration and service stand for the DSM.
The M2 hub, including the DSM, is mounted inside.
Figure 6. Electronic control boards for digital
signal processing for the DSM. Each board
controls 16 channels.
15
mechanical design of the lasers and opti-
cal tube assembly interfaces also re -
quires the teams full attention. Safety is
important when designing with 22 W
lasers as components and is taken very
seriously by the team. A consultant has
been contracted to provide support
to the safety analysis and much has been
learnt during the design of the current
4LGSF system.
The fnal design documentation for the
4LGSF is complete and the fnal design
review will take place in January 2011.
Procurement will then start for the opto-
mechanical and electronic system
assembly. A frst optical tube assembly
unit is expected for the summer of 2011
and the fnal design of the laser will be
reviewed in mid-2011.
ASSIST
The ASSIST test bench is a key com-
ponent of the Adaptive Optics Facility.
It will allow full characterisation of the
AOF in Europe before installation and
com missioning in Paranal. It also offers
stable, known and reproducible condi-
tions of turbulence (simulated by phase
screens) to establish whether the whole
assembly (wavefront sensors, real-time
computer, deformable mirror and the
algorithms implemented) meet the perfor-
mance specifcations.
The most striking feature of ASSIST is
that it requires a large 1.7-metre main
mirror to feed the light onto the M2 con-
vex deformable mirror. This mirror is
manufactured by AMOS (Belgium) and is
in the last stages of aspheric polishing
(see Figure 9). Acceptance is planned for
late 2010.
The bench consists of a tower with
the 1.7-metre mirror at the base and the
new generation M2 unit (hosting the
DSM) on top (see Figure 10). A system of
45-degree mirrors will feed artifcial
sources from the side. A star simulator
and turbulence generator feeds ASSIST
with sources simulating natural and
laser guide stars (with different focus
positions) and three phase screens
located at different altitudes. This setup
allows realistic testing of the AO modules
and complete testing of all the functio-
nalities. ASSIST will be delivered to
Garching in spring 2011 for integration in
the large tower with the mirrors well in
ad vance of the arrival of the DSM, planned
for early 2012.
Unit Telescope upgrade
In April 2008 it was recognised by the
project and upper management that the
scope of modifcations required to the UT
that will receive the AOF were substantial
and needed detailed planning and careful
Figure 7. Left: The ESO Raman fbre laser used to
test laser optics coatings for the launch telescope.
Right: The single arm Raman fbre amplifer from
MPB/TOPTICA showing the required 37.4 W of infra-
red 1178 nm power for the proper Na wavelength
output power (image from April 2010).
Lens L2 & mount
Main tube
Optical
base block
Field selector
mechanism
Lens L1 & quarter-
wave plate
Figure 8. The optical
tube assembly being
developed by TNO
Science & Industry.
Together with the beam
control and diagnostics
system this assembly
makes up the launch
telescope system for
the 4LGSF. Four similar
units are mounted on
the centrepiece of the
UT to launch the four
laser guide stars at
90 km altitude.
Figure 9. The ASSIST test bench (left) with the new
generation M2 unit on top of the structure. The large
1.7-metre mirror is at the bottom (hidden by the light
covers in this illustration) and the Nasmyth adaptor/
rotator test bench can be seen on the right-hand
side to allow the AO modules to be mounted. On the
right is the secondary mirror (AM2), a 140 mm
convex aspherical mirror manufactured by ASTRON,
that completes the main tower optics of ASSIST.
16
MUSE narrow-feld mode of GALACSI.
The ESO Optical Detector Team is us-
ing the fast readout OCam (developed by
LAOG, LAM and OHP; see Release
eso0922
2
) to develop an NGC controller
for the e2v CCD220 detector that will
be used for the WFS cameras. Fifteen
units have been ordered for SPHERE and
the AOF.
Project outlook
The year 2011 will be an exciting one dur-
ing which all the AOF project subsystems
will start assembly and testing. The frst
science thin shell mirror and the GRAAL
main assembly should be delivered dur-
ing the frst quarter of the year. GALACSI
integration will already be underway and
preliminary tests of the two AO module
subsystems will take place in the course
of the year. ASSIST will be delivered to
Garching and integrated during the sec-
ond half of 2011.
The AOF system test phase will start
when the DSM is delivered in the frst
quarter of 2012 and will proceed with
optical testing of the DSM, full system
tests of GRAAL and then GALACSI. This
will last for slightly more than one year.
The 4LGSF integration and test proceeds
in parallel to these activities and the frst
shipments to Paranal and the beginning
of the commissioning activities will take
place during the second half of 2013.
This shipment will trigger intense com-
missioning activities involving the whole
AOF team until the end of 2014, which
is the project goal for provisional accept-
ance Chile.
References
Arsenault, R. et al. 2010, Proc. SPIE, 7736, 0L
Baade, D. et al. 2009, The Messenger, 136, 20
Bonaccini Calia, D. et al. 2010, The Messenger,
139, 12
Feautrier, P. et al. 2010, Proc. SPIE, 7736, 0Z
Paufque, J. et al. 2010, Proc. SPIE, 7736, 1P
Stuik, R. et al. 2010, Proc. SPIE, 7736, 3M
Links
1
Deformable secondary mirror: http://www.eso.org/
public/announcements/ann1056/
2
OCam: http://www.eso.org/public/images/
eso0922b/
implementation. A specifc branch of the
AOF project was created to manage this
effort. At the AOF system fnal design
review, held in April 2010, a large number
of documents focused on this aspect and
received the approval of the review board
and the Paranal management. This re-
view constituted the kick-off for this sub-
project. The basis for the defnition of the
requirements is all the non-conformances
identifed in the Interface Control Docu-
ment between the AOF and the VLT. This
sub-project will implement: mechanical
interfaces between the UT and the
4LGSF electronics cabinets and launch
telescopes on the UT centrepiece (see
Figure 11 left); overhaul of the cooling
system and electrical supply; implemen-
tation of platform extensions to accom-
modate all the electronics cabinets;
modifcation to the Nasmyth guider arm
to accommodate the longer back focal
distance of GALACSIMUSE; def nition of
all cable (for power, signals, fbres, etc.)
routing in the UT; development and imple-
mentation of a new, easier to deploy,
M2 maintenance platform (see Figure 11
right); and specifcation of all the required
assembly, integration and testing facilities
needed in Paranal.
Complementary systems
There are a few important components
of the AOF that are being developed
as sub-contracts to other frms or to ESO
departments. The real-time computer
SPARTA for GRAAL and GALACSI is
managed by a sub-group of the AO De -
partment. The two identical real-time
computers will be delivered to the AOF
after SPHERE is delivered. SPHERE
is a single conjugate AO system using a
natural guide star on axis; the required
functionalities will be re-used by the
maintenance mode of GRAAL (for the
natural guide star on-axis). The GLAO
functionalities will build on this for the
GLAO mode of GRAAL and GALACSI.
These aspects will be completed by laser
tomography AO functionalities for the
Arsenault R. et al., Progress on the VLT Adaptive Optics Facility
Figure 11. Left: Concept of a Nasmyth platform
extension for the 4LGSF electronic cabinets and
heat exchangers below the existing platform. Right:
One of the two concepts for a deployable M2 main-
tenance platform to enable quick maintenance work
on the new generation M2 unit.
Figure 10. The main
1.7-metre mirror of
ASSIST is shown at
AMOS in Belgium
during the last step of
aspherical polishing.
17
mentation section page, p. 4); the remain-
der are partly equipped and under test or
in the process of being outftted with
ALMA front-end and back-end systems
at the Operations Support Facility (OSF).
Many more antennas are in various
stages of assembly in the contractor
camps at OSF. Among these are the frst
European antennas from the AEM con-
sortium; fve of these have been fully
assembled at the time of writing (see Fig-
ure 3) and the frst ones are undergoing
performance tests and tuning with the
goal of being delivered to the ALMA pro-
ject in early 2011.
Initial results of the all-sky pointing tests
have been excellent with the root mean
square on pointing accuracy below 1 arc-
second (see Figure 4 for an example of a
pointing run). The surface accuracy of the
European antenna tested so far is also
excellent. The panels are mounted and
aligned in the assembly procedures to a
good initial accuracy. Preliminary tests
on the frst antenna have shown that, fol-
lowing a partial initial cycle of holography
measurements and panel adjustments,
the surface can be brought within the
ALMA specifcations (see example in Fig-
ure 5). Testing of many aspects, such as
the dependence of the antenna perfor-
mance on environmental conditions and
of the fnal fgure and stability of the sur-
face ac curacy, are starting at the time of
writing.
Leonardo Testi
1
Richard Hills
2
Robert Laing
1
Stefano Stanghellini
1
Wolfgang Wild
1
1
ESO
2
Joint ALMA Observatory
We report on the status and progress
of the ALMA project and the expected
timeline and capabilities for Early Sci-
ence. Over the past year, the progress
on ALMA construction and on the com-
missioning activities has been huge.
At the time of writing the observatory is
progressing on the initial phases of sci-
ence verifcation and preparing to open
to external users to begin Early Science
observations.
The year 2010 has seen extraordinary
progress in the ALMA project. At the
be ginning of the year, with three fully
equipped antennas working as an inter-
ferometer, Commissioning and Sci-
ence Verifcation (CSV) activities offcially
started at the Array Operations Site
(AOS) on the Chajnantor plateau at an
elevation of 5000 metres (Testi, 2010).
Under the leadership of Richard Hills and
Alison Peck at the Joint ALMA Offce,
the CSV team has been steadily working
to test and improve the performance
of the whole ALMA system. This process
has culminated in the production of the
frst ALMA test images in the second part
of the year (see Figure 1 for an early
example). At the time of writing the CSV
team is working to obtain science dem-
onstration data with the eight-antenna
interferometer at the high site and is
focusing on refning the calibration plan,
in addition to continuing the commis-
sioning activities on the equipment that
is constantly being delivered.

On the hardware construction side, there
has been steady progress throughout
the year thanks to the tremendous effort
from the project teams in all regions
and the work of the Assembly Integration
and Verifcation group in Chile. In par-
ticular, focusing on deliverables from the
European side, the ALMA front-end
cartridges for Band 7 (275373 GHz, pro-
duced at IRAM in France) and Band 9
(602720 GHz, produced by NOVA in
the Netherlands) as well as the water
vapour radiometers (produced by the
Swedish company Omnisys) are in a very
advanced stage of production and more
than half of the total number of units
have been delivered to the project. The
test data shown in Figure 2 were ob -
tained with the Band 7 and Band 9 car-
tridges. Phase correction using the water
vapour radiometers (see Nikolic et al.,
2009) has been demonstrated as part of
commissioning; this represents a key
step towards validating the ALMA calibra-
tion strategy for high frequency and high
angular resolution observations.
Naturally, the most visible pieces of
hardware of the ALMA system are the
antennas. ALMA has provisionally
accepted fourteen antennas. Eight anten-
nas are fully equipped and working as
an interferometer at the high site (see the
image on the Telescopes and Instru-
ALMA Status and Progress towards Early Science
Figure 1. ALMA test data obtained in Bands 3 and 9
on the nearby star-forming galaxy NGC 253. The
large image is a VISTA infrared image of NGC253
(see ESO PR 1025). The data shown is the CO(10)
integrated emission; at right the CO(65) integrated
emission contours are overlaid on the 0.45 mm con-
tinuum emission.
NGO 253 Band 9 cont|nuum, OO J=6-5 contours
00
h
47
m
33
s
.7 33
s
.5 33
s
.3 33
s
.1 32
s
.9 32
s
.7
-251724
22
20
18
16
14
A|MA/8 OO J=1-0 lntegrated lntens|ty
A
|
M
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

tttttttttttt
eeeeeeeeeeeeeeeeeeee
ssssssss
tttttttt

dddddddd
aaaaaaaa
ttttt
aaaaa
00
h
47
m
35
s
34
s
33
s
32
s
31
s
-251700
40
50
30
20
10
50
40
C
r
e
d
i
t
:

A
L
M
A

C
S
V

t
e
a
m
18
Figure 2. ALMA test data in Band 7 of the molecular
outfow in the star-forming region NGC1333. The red
and blue contours show the high velocity CO(32)
emission while the 0.85 mm continuum emission is
shown as greyscale.
IRAS4B contin & hv CO
Right Ascension (J2000)
03
h
29
m
13
s
.5 13
s.0
12
s
.5 12
s
.0 11
s
.5
311300
56
52
04
08
12
16
20
A
L
MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

ttttttttttttttttttttttttttttttttttt
eee
s
ttt

dddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
ttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
C
r
e
d
i
t
:

A
L
M
A

C
S
V

t
e
a
m
Figure 3. The AEM antenna camp at the OSF. Five
antennas are fully assembled while a sixth is inside
the temporary shelter on the right ready for the
installation of the main refector assembly; parts of
the seventh antenna are also being assembled.
C
r
e
d
i
t
:

E
S
O
/
A
E
M

c
o
n
s
o
r
t
i
u
m
dS
N N
A
+2
-2
E W
dE
N N
A
+2
-2
E W
dv
N N
A
+5
-5
E W
AM
N N
A
+5
-5
E W
Z
dS
+2
-2
80
Z
dZ
+2
-2
80
|
2.0
0.8
R
arcsec
S
R
D
Z
0.5
arcsec
b|ns
|ast observat|on 23 October 2010
W
N
1
Figure 4. Example of all-sky optical
pointing session for the frst AEM
antenna. The top panels show the
pointing errors as a function of
azimuth and elevation of the observed
stars. The bottom right circular
diagram shows the position of the
observed stars on the sky. The bottom
left cir cular panel shows the point-
ing error distribution, which in this
case corresponds to an all sky rms
accuracy of 0.8 arcseconds.
Testi L. et al., ALMA Status and Progress towards Early Science
19
During the autumn of 2010, the ALMA
project has been going through a series
of external reviews that have focused
on the progress and current results of the
CSV activities and the readiness for oper-
ations, culminating in the ALMA Annual
External Review. These reviews scruti-
nised all the aspects of the ALMA project
and provided advice to the ALMA Direc-
tor and the ALMA Board on the status
and progress of ALMA construction. The
main focus of the reviews has been to
assess the readiness of ALMA for Early
Science. The conclusion of the reviews is
that hardware delivery, system verifca-
tion and commissioning, as well as the
deployment of all the procedures for the
start of science operations, are all pro-
gressing at the pace required to start
Early Science observations in the second
half of 2011. Nevertheless, the opening
of the observatory to external users will
result in additional challenges to meet the
standards expected for ALMA data and
run science operations at the same time
as construction and commissioning. For
these reasons the ALMA science advi-
sory committees have all recommended
that the main focus remains on the full
ALMA construction and operations and
that the Early Science capabilities are
focused on a limited set of well-tested
modes.
The announcement of opportunity for
Early Science observations will be re -
leased in the frst quarter of 2011 with the
deadline for proposal submission before
the summer. The precise capabilities that
will be available for Early Science will be
announced as part of the call for propos-
als, but it is likely that the array will be
offered in two compact confgurations
of 16 antennas (with the maximum base-
line likely to be of the order of 250 metres).
The limited instantaneous coverage of
the uv-plane will imply that most projects
will require Earth rotation synthesis to
achieve good imaging capabilities during
Early Science. The available observing
modes will be limited to single feld inter-
ferometry using Bands 3, 6, 7 or 9
(84116 GHz, 211275 GHz, 275373 GHz,
602720 GHz, respectively), full polari-
sation capabilities and a range of single
spectral resolution correlator modes.
During the frst year of operations, scien-
tifc observations will be limited to a frac-
tion of the total available time, not ex -
ceeding 30%, as the frst priority of the
project will be to work on completing and
commissioning the full ALMA system.
A series of workshops, schools and
tutorials have been and are being organ-
ised to prepare European astronomers
for ALMA. The European ALMA Regional
Centre (ARC) and its nodes located in
several European countries have put in
place a strong effort to support the com-
munity uptake of Early Science. On 6
and 7 April 2011, the European ARC will
organise a tutorial at ESO Headquarters
in Garching focused on the use of the
ALMA software for Early Science (see the
full announcement on page 49). In par-
ticular, the tutorials will focus on the use
of the ALMA observing tool, which will
be used by ALMA users for the prepara-
tion of observing proposals and to set up
the approved projects for observation at
the telescope.
References
Nikolic, B. et al. 2009, The Messenger, 131, 14
Testi, L. 2010, The Messenger, 139, 52
Figure 5. Adjusting panels on the AEM antennas
using the manual panel adjuster tools. Note that this
operation can be performed walking on the refector
surface when the antenna is stowed; the hole on the
dish surface to the rear of the image is used for the
optical pointing telescope.
Figure 6. Example results of a holography run after
tuning the position of approximately half of the pan-
els on the AEM dish. In this particular case the sur-
face was found to be set to approximately 15 m rms.
C
r
e
d
i
t
:

E
S
O
/
A
E
M

c
o
n
s
o
r
t
i
u
m
A
s
t
r
o
n
o
m
i
c
a
l

S
c
i
e
n
c
e
Colour image of the nearby intermediatelow
mass star formation region R Coronae Australis
taken with the MPG/ESO 2.2-metre telescope.
Images through B-, V- and R-band flters were
combined. The star in the centre is R Coronae
Australis itself (an Herbig Ae/Be star), which is sur-
rounded by an embedded cluster of young stars,
the Coronet. Dark clouds and refection nebulosity
from illuminated dusty molecular clouds encircle
the region.
21
Astronomical Science
Andreas Seifahrt
1, 2, 3
Hans Ulrich Kuf
3
Gnther Zngl
4, 5

Jacob Bean
2, 6
Matthew Richter
1

Ralf Siebenmorgen
3
1
University of California at Davis, USA
2
Institut fr Astrophysik, Gttingen,
Germany
3
ESO
4
Meteorologisches Institut der LMU,
Mnchen, Germany
5
Deutscher Wetterdienst, Offenbach,
Germany
6
HarvardSmithsonian Center for
Astrophysics, Cambridge, USA
Ground-based astronomical observa-
tions suffer from the disturbing effects
of the Earths atmosphere. Oxygen,
water vapour and a number of atmos-
pheric trace gases absorb and emit
light at discrete frequencies, shaping
observing bands in the near- and
mid-infrared and leaving their fnger-
prints telluric absorption and emis-
sion lines in astronomical spectra.
The standard approach of removing the
absorption lines is to observe a telluric
standard star: a time-consuming and
often imperfect solution. Alternatively,
the spectral features of the Earths
atmosphere can be modelled using a
radiative transfer code, often deliver-
ing a satisfying solution that removes
these features without additional obser-
vations. In addition the model also
provides a precise wavelength solution
and an instrumental profle.
The Earths atmosphere consists of a
rich gas mixture. While its main constitu-
ent, nitrogen (N
2
), does not exhibit any
rotationalvibrational transitions, most
other molecules do. Strong absorption
line systems from water vapour (H
2
O),
carbon dioxide (CO
2
) and ozone (O
3
)
shape the well-known photometric band-
passes in the near- and mid-infrared.
In addition, oxygen (O
2
) shows strong ab -
sorption bands in the red optical; other
molecules such as nitrous oxide (N
2
O),
carbon monoxide (CO), or methane (CH
4
)
contribute noticeably to the atmospheric
transmission losses in the near- and mid-
infrared, often hampering the observation
of important astrophysical lines.
While the Earths atmosphere absorbs
light from astronomical sources at a large
number of frequencies from the ultra-
violet to radio wavelengths, it also emits
light in the same transitions, radiating
its thermal energy into space. At wave-
lengths longer than about 2300 nm, the
emission originating from the Earths
atmosphere competes with the signal
from astronomical sources. The break-
even point depends on many factors,
such as the spatial resolution, slit width,
and detector characteristics of the
spectrograph, but also on the height of
the observing site. Since the atmos-
pheric emission is spatially extended, it
affects the spectrum of the astronomical
target similarly to the nearby sky and
can thus be removed by beam-switching
(nodding) techniques or by ftting and
subtracting the two-dimensional signal
along the slit.
Empirical calibration
Removing the telluric absorption features
is more diffcult since their exact signa-
ture is only imprinted in the source spec-
trum. Traditionally, telluric absorption
features are removed by observing a so-
called telluric standard star before or
after the observation of the science target
at the same airmass, subsequently divid-
ing the spectrum of the science target
by that of the telluric standard. Typically,
early-type stars with spectral types rang-
ing from early B to late A are used for this
purpose, since they exhibit rather fea-
tureless spectra, except for strong hydro-
gen lines. Since these stars are often
fast rotators, other weak stellar features,
for example helium lines, are further
suppressed. Alternatively, early-to-mid
G-type stars can serve as telluric stand-
ards as well, since high resolution Fourier
transform spectra (FTS) of the Sun pro-
vide the necessary template to compen-
sate for their intrinsic stellar features.
The observation of telluric standard stars
is a standard procedure and commonly
employed for all spectrographs operating
on ESO telescopes. It is often part of
the calibration plan for the instrument
and, in these cases, is automatically pro-
vided for service mode observations.
Since telluric lines generally do not scale
linearly with airmass and observing
conditions are time variable, it is neces-
sary to observe a telluric standard star at
the same airmass and close in time to
the science target in the same instrumen-
tal setup. Special software tools at the
telescope allow the effcient selection of
an appropriate standard star.
Nevertheless, using standard stars as
empirical calibrators has several disad-
vantages. The observation of standard
stars is time-consuming, especially when
science targets are bright or high signal-
to-noise requirements are to be met.
For the brightest standard stars, the time
required for telescope and instrument
presets ultimately limits the effciency of
this method. Moreover, on instruments
like CRIRES, the instrumental profle
depends on the performance of an adap-
tive optics system, and thus, on the
observing conditions and source bright-
ness. Changes in the instrumental pro-
fle lead to changes in the line shape of
unresolved spectral lines. In these cases
standard stars rarely provide a perfect
match to the science target, effectively
limiting the precision with which telluric
features can be removed.
Other shortcomings originate from the
intrinsic stellar features of the standard
stars. Hydrogen and helium lines in
early-type standards cannot be perfectly
removed and thus affect the line profles
of the same species in the spectrum
of the science target. A further complica-
tion in removing these lines occurs
when the wavelength coverage of the
spectrum is shorter than the line widths
of broad hydrogen lines, which can eas-
ily be the case for CRIRES. Early-type
stars can also exhibit other spectral fea-
tures, such as oxygen or carbon lines
in the near-infrared, often as emission
features originating under non-local ther-
modynamic equilibrium (non-LTE) con-
ditions. Similarly, mismatches in the line
depths of solar-type standard stars
with the solar FTS atlas, attributable to
abundance differences or deviating effec-
tive temperatures, can leave residuals
of the standard stars intrinsic features in
the fnal spectrum.
Precise Modelling of Telluric Features
in Astronomical Spectra
22
Modelling telluric features
Alternatively, the telluric absorption spec-
trum can be synthesised, using a radia-
tive transfer code in combination with a
layered model of the Earths atmosphere
and a database containing the transition
data for all molecules in consideration.
We have successfully used the program
LBLRTM (Clough et al., 1992) for this
purpose. LBLRTM is a non-commercial
layer-by-layer radiative transfer code
tailored to produce telluric spectra under
various atmospheric geometries. The
code is generally available to the commu-
nity and uses the high resolution trans-
mission molecular absorption database
HITRAN (Rothman et al., 2009) as a
molecular line database. HITRAN con-
tains energy levels, frequencies, line
strengths, pressure-broadening and -shift
coeffcients, for more than 1.7 million
spectral lines of 42 different molecules
and their common isotopes from the red
optical to the sub-mm.
A layered model of the Earths atmosphere
serves as the primary input for LBLRTM
and contains temperature, pressure
and molecular abundance information as
a function of atmospheric height at the
observatory. For a representative, yet
accessible model of the atmospheric
conditions at the time of the observation,
we supplement a static model of the
atmosphere with meteorological data for
temperature, pressure and humidity of
the troposphere and lower stratosphere
(surface height 26 km). LBLRTM
converts the atmospheric model into indi-
vidual isothermal layers and calculates
absorption and emission spectra for a
given path through the atmosphere, i.e. for
a given zenith angle or airmass.
A single model run takes only a few sec-
onds on a standard desktop PC. The
re sulting spectrum has an intrinsic reso-
Figure 1. Examples of synthesised telluric spectra
ftted to CRIRES observations of standard stars.
Each panel shows the measured spectrum in black
and the model overplotted in red. We show the
residuals after division by the models, i.e., after
removal of the telluric absorption lines, at the bottom
of each panel. Top panel: Water vapour (H
2
O) lines
in the H-band. Middle panel: Nitrous oxide (N
2
O) and
water vapour (H
2
O) in the L-band. Bottom panel:
Ozone (O
3
), carbon monoxide (CO), carbon dioxide
(CO
2
), and water vapour (H
2
O) lines in the M-band.
1502 1503 1504 1505 1506 1507 1508
Wave|ength (nm}
1.00
0.95
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1.10
0
/
O
N
o
r
m
a
|
|
s
e
d

f
u
x
1.05
1502 1503 1504 1505 1506 1507 1508
3900 3905 3910 3915
Wave|ength (nm}
1.00
0.95
0.20
0.40
0.60
0.80
1.00
0
/
O
N
o
r
m
a
|
|
s
e
d

f
u
x
1.05
3900 3905 3910 3915
4695 4700 4705 4710
Wave|ength (nm}
1.0
0.9
0.20
0.00
0.40
0.60
0.80
1.00
0
/
O
N
o
r
m
a
|
|
s
e
d

f
u
x
1.1
4695 4700 4705 4701
Seifahrt A. et al., Precise Modelling of Telluric Features in Astronomical Spectra
23
lution that is set to resolve the narrowest
telluric lines in a given spectral region.
Resolving powers are thus often of the
order of 10
6
and the model spectrum
needs to be convolved with the instru-
mental profle of the spectrograph to
match the measured telluric spectrum.
To frst order, the instrumental profle can
be represented by a single Gaussian
function. More complex profles might be
used if higher precisions and signal-to-
noise ratios need to be achieved. At the
same time, the abundance of certain
molecules might be over- or underesti-
mated by the atmospheric model, espe-
cially for strongly variable species such
as water vapour. Given the extreme geo-
graphical location of most astro nomical
observatories, including Cerro Paranal,
general meteorological models rarely
reproduce the exact water vapour levels
on site. Thus, the model data need to
be matched to the actual observing con-
ditions. A chi-squared ftting algorithm
can be employed to solve for the instru-
mental profle and to re-adjust the abun-
dance of the molecular species in the
atmospheric model while minimising the
residuals between the ob served spec-
trum and the spectral model.
We show some example fts to spectra
obtained with the CRIRES spectrograph
in Figure 1. The resolving power was
R~6500080000 as measured from the
width of the instrumental profle, which
was ftted at the same time as a single
Gaussian function. More examples and a
detailed discussion of the ftting per-
formance can be found in Seifahrt et al.
(2010). In all cases, the abundance of
most atmospheric species was slightly
modifed by a few percent during ftting to
achieve an optimal result. The ft pro-
cess also included the wavelength solu-
tion for the CRIRES spectra. This step
is critical, since narrow and steep telluric
lines are sensitive to small wavelength
errors and produce notable residuals for
wavelength mismatches as small as
~30m/s. A typical ft process, including
various runs of LBLRTM takes about
two minutes to converge over the spec-
tral coverage of a single CRIRES detec-
tor. We have also successfully applied
the code to optical high resolution spec-
tra from UVES and to low resolution near-
infrared spectra from SINFONI.
Wavelength calibration as added value
Fitting telluric lines to empirical spectra
not only provides a model of the telluric
absorption, it also delivers a precise
wavelength solution. This is an important
added value of this method, given that
the wavelength calibration for high reso-
lution near- and mid-infrared spectra is
often challenging. Rare gas emission line
lamps used in low resolution near-infrared
spectrographs, such as He, Ne, Xe, and
Kr lamps, provide only a sparse line den-
sity and are only of very limited use at
high spectral resolution (see also Aldenius
et al., 2008). Even the rich spectra of
ThAr emission line lamps, commonly em -
ployed by high resolution optical spec-
trographs, have a much lower line density
in the near-infrared than in the red opti-
cal. Typical line densities of ~400 lines
per 100nm around =1000nm drop
quickly to less than ~20 lines per 100nm
at =2500 nm. Given the typical wave-
length coverage of ~/200 per CRIRES
detector, many spectral settings remain
poorly calibrated. Moreover the dynamic
range of the ThAr emission spectrum is
very high due to the strong contrast be -
tween the Th and Ar lines, leaving weaker
lines at low signal-to-noise ratios while
nearby stronger lines quickly saturate.
Often atmospheric OH* emission lines
(commonly referred to as airglow emis-
sion) are used for wavelength calibration
in the near-infrared, especially for low-
and mid-resolution spectrographs. OH*
is a chemical radical and its transitions
originate from non-LTE processes in high
atmospheric layers. They appear in emis-
sion but not in absorption in astronomi-
cal spectra and constitute the dominant
background source in the near-infrared
J- and H-bands. However, neither the line
density nor the line strength of individual
OH* lines is high enough to make this
species useful for the wavelength calibra-
tion of high-resolution spectrographs.
For example, the small spatial pixel scale
of CRIRES (86 milliarcseconds/pix) makes
the use of OH* lines especially challeng-
ing, given that each pixel covers less than
10
2
arcsecond
2
on the sky.
In contrast, the atmospheric features of
other regular molecules in LTE, as we
model them with LBLRTM, show a high
density of strong absorption lines and
thus provide a natural in situ wavelength
calibration. Since they are imprinted
in the source spectrum before the light
reaches the spectrograph, these lines
suffer the same instrumental effects as
the intrinsic lines in the stellar spectrum.
Hence, in contrast to all emission line
sources used to wavelength calibrate
long-slit spectra, telluric absorption lines
provide an intrinsically more precise
calibration source, especially since lamp
emission lines are recorded in separate
exposures, often hours after the science
spectra and thus after potential changes
in the spectrograph setup.
Performance and limitations
As can be seen in the examples pre-
sented in Figure1, the modelling of tellu-
ric absorption lines is not perfect, but
leaves residuals at the 2% level for most
lines when observed at high spectral res-
olution. A few lines exhibit even stronger
residuals. The main reasons for these
mismatches are imprecise line data in the
HITRAN database and insuffcient treat-
ment of line coupling (also known as line
mixing) in the radiative transfer code.
The latter is, for example, responsible for
the strong residuals of the CO
2
Q-branch
at 4695 nm (see the lower panel of Fig-
ure1). Other limitations may arise from
uncertainties in the determination of the
instrumental profle and the atmospheric
model.
Line data in HITRAN have strongly vary-
ing accuracy levels. Typical uncertain-
ties of line positions range from a few
to several hundred m/s, but can be as
high as several km/s in extreme cases.
Line strengths are rarely precise to the
1% level. However, the HITRAN database
is constantly updated and the data qual-
ity will further improve in the future. Also,
individual line data can be re-ftted when
comparing synthesised spectra with
high resolution spectra of standard stars
before applying the code to spectra of
science targets.
Despite these limitations, synthesised tel-
luric spectra can provide the same or
even better telluric correction than empir-
ical spectra. The question of when to
use empirical spectra or a telluric model
depends on a number of factors and
24
ard star observations would have com-
promised the achievable long-term radial
velocity precision, strongly favouring the
telluric modelling approach.
Last but not least, telluric models are
not only used to wavelength calibrate and
to correct for the telluric features in ob -
served spectra, but also to plan observa-
tions and predict the atmospheric
throughput based on meteorological fore-
casts. For example, the exposure time
calculator for CRIRES employs a telluric
model to allow the user to check spectral
settings for the impact of telluric absorp-
tion under different water vapour levels.
References
Aldenius, M. et al. 2008, The Messenger, 133, 14
Clough, S. A., Iacono, M. J. & Moncet, J.-L. 1992,
J. Geophys. Res., 97, 15761
Rothman, L. S. et al. 2009, Journal of Quantitative
Spectroscopy and Radiative Transfer, 110, 533
Seifahrt, A. et al. 2010, A&A, in press, arXiv:1008.3419
Bean, J. L. et al. 2010a, ApJ, 713, 410
Bean, J. L. et al. 2010b, The Messenger, 140, 41
Lebzelter, T. et al. 2010, The Messenger, 139, 33
should be decided on a case-by-case
basis. However, the use of a telluric
model should be strongly considered for
regions where standard stars have spec-
tral features, when instrumental setups
(including the instrumental profle) cannot
be reproduced, or when observations
encompass a wide range in airmass.
Some recent and ongoing science pro-
grammes entirely rely on telluric models
for calibration. One example is the
CIRESPOP project, dedicated to pro-
viding the community with a high reso-
lution, high signal-to-noise near-infrared
spectral atlas of several stars across
the HertzsprungRussell diagram
(Lebzelter et al., 2010). Data reduction
for this project employs telluric models
to wavelength calibrate settings that
lack coverage by ThAr lines or gas cells.
CRIRESPOP concentrates on very
bright stars and uses a large number of
CRIRES settings. Obtaining spectra
of telluric standard stars for each science
target would be very time consuming,
at least doubling the total observing
time. Most importantly, the programme
in cludes observations of early-type stars
to map their spectra to search for unac-
counted lines. This process would prove
impossible if a standard star of similar
spectral type was used as an atmospheric
calibrator.
Another project utilising telluric models
is the CRIRES radial velocity campaign
to search for extrasolar planets around
mid-to-late M-type stars (Bean et al.,
2010a,b). The wavelength calibration for
this project relies on a custom-built am -
monia gas cell, given the need to deter-
mine radial velocities to the m/s level.
The telluric features present in all obser-
vations obtained by this programme
need to be corrected to a high level of
precision. Obtaining high signal-to-noise
standard star spectra for all science
observations would have made the pro-
ject less time-effective. Also, the instru-
mental profle between the science
observation and the standard star must
match; a challenging requirement for
CRIRES. Given the sensitivity of the radial
velocity method to small changes in
the instrumental profle, the use of stand-
Seifahrt A. et al., Precise Modelling of Telluric Features in Astronomical Spectra
NGC 3603, shown here in a VLT FORS three-colour
composite (from V-, R- and I-band images), is a
young Galactic starburst cluster situated at a dis-
tance of 6.8 kiloparsecs. With many dozens of mas-
sive young hot stars and an age of a few Myr, it is
the focus of many studies as a nearby template for
extragalactic starburst environments.
25
Michael Sterzik
1
Stefano Bagnulo
2
Armando Azua
3
Fabiola Salinas
4
Jorge Alfaro
4
Rafael Vicuna
3

1
ESO
2
Armagh Observatory, United Kingdom
3
Department of Molecular Genetics and
Microbiology, Pontifcia Universidad
Catlica de Chile, Chile
4
Faculty of Physics, Pontifcia Universi-
dad Catlica de Chile, Chile
Homochirality, i.e., the exclusive use of
L-amino acids and D-sugar in biologi-
cal material, induces circular polarisa-
tion in the diffuse refectance spectra of
biotic material. Polarimetry may there-
fore become an interesting remote
sensing technique in the future search
for extraterrestrial life. We have explored
this technique and performed a labora-
tory experiment making an exotic use
of an astronomical instrument. During a
period when EFOSC2 was detached
from the Nasmyth focus to host a visitor
instrument at the NTT, we have ob-
served various samples of biotic and
abiotic material and measured their
linear and circular polarisation spectra.
Among the various targets, we have in-
cluded samples of the hypolithic cyano-
bacteria species Chroococcidiopsis
i solated from the Coastal Range of the
Atacama Desert. To our knowledge,
these are the frst and highest preci-
sion measurements of circular polarisa-
tion using living material and obtained
with an astronomical instrument.
Motivation
The building blocks of life are chiral. Their
molecular structure lacks an internal
plane of symmetry, and their mirror image
cannot be superimposed on their origi-
nal image. The term chirality is specif-
cally used when a molecule (or an object)
exists in both mirror-symmetric confgu-
rations. The human hands are the classic
example that illustrates the concept of
chirality, and the term chiral itself comes
from the Greek word for hand, .
In chemistry, the two images of a chiral
molecule are called enantiomers, and the
two forms are generally referred to as
right-handed and left-handed, or dextro-
rotatory and levorotatory.
The term homochirality is used when a
molecule (or a crystal) may potentially
exist in both forms, but only one is actu-
ally present. Homochirality character-
ises life as we know it: all living material
on Earth contains and synthesises sugars
and nucleic acids exclusively in their
right-handed form, while amino acids and
proteins occur only in their left-handed
representation. However, in all these
cases, both enantiomers are chemically
possible and energetically equal. The
reasons for homochirality in living mate-
rial are unknown, but they must be
related to the origin of life. It is still dis-
puted whether bioactive molecules (and
with them small enantiomeric excesses)
were delivered to Earth (e.g., by mete or-
ites) or whether (pre-)biotic chemistry
started on Earth (Bailey et al., 1998).
Undoubtedly, however, chemical and bio-
logical processes on Earth must have
favoured the selection of one-handed
biomolecules leading towards homochi-
rality. If similar evolutionary scenarios
naturally occur elsewhere in the Universe,
homochirality may be a universal hallmark
of all forms of life.
Chirality induces optical activity: each
enantiomer rotates the refected (or trans-
mitted) light in opposite directions, and
homochirality guarantees that there will
be an excess of circularly polarised light
in one direction. This opens up the inter-
esting possibility that biosignatures could
be sensed remotely by means of polari-
metric techniques.
This chain of arguments, which had been
raised in various articles in the scientifc
literature (e.g., Wolstencroft et al., 2004),
was also discussed during the workshop
Astrobio 2010
1
held in Santiago de Chile
last January. For the frst time, an inter-
national and interdisciplinary conference
that aimed to cover major topics in astro-
biology was organised and hosted in
Chile. The topics covered included the
origins of life, the chemistry of the Uni-
verse, extrasolar planetary systems, and
the search for life in the Solar System.
A prominent topic of discussion was the
Atacama Desert as an example of an
extreme environment on Earth, and the
microbial colonisation of subsurface lay-
ers in halites (rock salt) and quartz rocks
by specifc cyanobacteria. In the most
hostile environments (exceptional aridity,
salinity, and extreme temperatures), a
primitive type of cyanobateria, Chroococ-
cidiopsis, can be the sole surviving
organism. This has interesting implications
for the potential habitability, and eventual
terraforming, of certain areas on Mars
(Friedmann & Ocampo-Friedmann, 1995).
The idea of using the ESO Faint Object
Spectrograph and Camera (EFOSC2)
to investigate samples of Chroococcidi-
opsis extracted from the underside of
Atacama Desert quartzes and to measure
their circular polarisation in a laboratory
experiment arose at this conference.
The idea looked appealing, because only
limited, and sometimes contradictory,
reports about circular polarisation meas-
urements of biotic material as a signature
of homochirality are available in the litera-
ture. Moreover, the successful use of
an astronomical instrument for the frst
time for that purpose could serve as a
benchmark for further applications of this
method in astrophysics.
A detailed feasibility study as well as
the production of signifcant quantities of
Chroococcidiopsis were prepared and
in itiated in the following weeks. Since
there is no formal process in place to ob -
tain observing time for laboratory
experiments, the ESO Director General
was asked for authorisation. He approved
the experiment under the condition that
it did not pose any risk to the instrument.
This could be achieved by using EFOSC2
when it was not attached at its nominal
Nasmyth focal station (i.e. during an ex -
tended visitor instrument run), but keep-
ing it in a horizontal position (which avoids
the possibility of any material falling on
the entrance window). Our laboratory
is shown in Figure1, left. EFOSC2 is
de tached from the New Technology Tele-
scope (NTT). A microphotograph of
our main target, Chroococcidiopsis, is
displayed on the right.
Experiment
During one week in June 2010, three
of us (Pontifcia Universidad Catlica
Astronomy Meets Biology:
EFOSC2 and the Chirality of Life
26
In the Philodendron leaf we detected
both broad polarised features (with an
amplitude of ~0.5%), and, superposed
on them, a narrow feature at about
680nm (with an amplitude of ~0.05%
over the continuum), both well above
the statistical noise (the green line shows
the null profle). The behaviour of Stokes
V in the continuum closely follows the
refectivity, shown with a solid black line
(this is known as the Cotton effect).
This behaviour closely resembles the dif-
fuse refectance circular dichroism
spectra of leaves as seen by Wolstencroft
et al. (2004). The narrow feature around
680nm is very similar to the results
shown by Gregory & Raps (1974) in their
transmission spectroscopy of chloro-
plasts and is related to the chlorophyll-a
pigment response.
The interpretation of the results for
Chroococcidiopsis is the most interesting.
In the continuum, Stokes V shows a
behaviour similar to that observed for the
white screen feld, and hence is likely
to be of instrumental origin. A number of
narrow, low amplitude features appear
superposed on the continuum. In order to
prove that these features are real (and
not spurious signals, e.g., caused by the
CCD readout noise) we also obtained
ultra-high signal-to-noise ratio measure-
ments with the retarder waveplate at 0,
90, 180 and 270. With these retarder
waveplate settings, we would theoreti-
cally expect a null signal. The resulting
profle, shown with the red line in Fig-
ure3, and arbitrarily rectifed to zero for
display purposes, shows no high fre-
quency signatures. This curve, together
with a consistently fat null profle (shown
student Fabiola Salinas, Stefano Bagnulo
and Michael Sterzik) were busy using
EFOSC2 to observe samples of minerals
and paints (quartz, salt, sugar, white
fat-feld screen), leaves (Philodendron,
Ficus, Scheffera) and cyano bacteria flms
deposited on flter paper. All samples
were prepared as thin sheets inserted in
the fat-feld screen position. An inte-
grating sphere (the usual calibration lamp
mechanism for EFOSC2 when attached
to the 3.6-metre telescope) was used
for diffuse illumination. EFOSC2 covers
large spectral regions in the spectropola-
rimetry mode: we used mostly grism
13 to cover the range 370930nm, with
a spectral resolution of ~2.3nm (we
adopted a 1-arcsecond slit width). Both
linear and circular polarisation measure-
ments of the samples were obtained,
using the /2 and /4 retarder wave-
plates, respectively (we note that the /4
retarder waveplate is a recent addition to
the instrument, see Saviane et al. [2007]).
Polarimetric measurements were taken
by combining several pairs of expo-
sures obtained with different position
angles of the retarder waveplates (45,
45, 135, 225 for circular polarisation,
and 0, 22.5, 45, ... 335.5 for linear
polarisation measurements, measured
from the principal plane of the Wollaston
prism). This beam swapping technique
minimises the instrumental effects
for a detailed description of the measur-
ing strategies, see, e.g., Bagnulo et
al. (2009). For inorganic samples and for
leaves, the typical observation cycles
lasted 20 minutes, dominated in practice
by overheads (readout time and retarder
waveplate setting), and allowed us
to measure the polarisation level with an
error bar of the order of 10
4
per spectral
bin. Measurements of the cyanobacte-
ria lasted several hours, but for a total
integration time of just about 10 minutes.
This exposure allowed us to reach an
error bar of 10
6
per spectral bin. These
fgures refer only to the statistical errors
due to Poisson noise. With our ultra-high
signal-to-noise ratio measurements,
we have certainly hit the limits imposed
by the polarimetric optics and the ex -
perimental conditions we will come
back later to this important point. Here we
note that the reliability of the error bars
(in terms of statistical error) has been vali-
dated with the use of so-called null-
profles (i.e. the difference between Stokes
profles obtained from different pairs of
exposures). Null profles were found scat-
tered around zero within the error bars.
Results and outlook
Figures2 and 3 show the results
ob tained from our circular polarisation
measurements (Stokes V normalised
to the intensity) of a Philodendron leaf,
and a flm of Chroococcidiopsis depos-
ited on flter paper, respectively. We
also obtained calibration measurements
of a white screen fat-feld, produced
by a barium sulphate based white refec-
tance coating. The circular polarisation
measured for the screen fat-feld was
found as a negative continuum of about
0.05%, with the absolute value slowly
increasing towards shorter wavelengths.
No narrow features or signals are pre-
sent in the polarisation spectra of this ref-
erence source.
Figure 1. Left: The EFOSC2 instru-
ment is shown detached from the
NTT. The instrument attached to the
Nasmyth focus (on the left) is ULTRA-
CAM. Right: A microphotograph of
the cyanobacteria Chroococcidiopsis,
enlarged 100 times.
Sterzik M. et al., Astronomy Meets Biology: EFOSC2 and the Chirality of Life
27
with the green line offset to +0.02% in
the right panel) suggests that the narrow
features are indeed real. We notice
again the prominent chlorophyll-a feature
around 680nm, but may also infer sig-
natures of other pigments like carotinoids
and phycocyanins. Our measurements
also appear consistent with polarisation
spectra obtained from a different type of
cyanobacteria by Sparks et al. (2009).
While we kept the noise confdently at
a very low level, the polarimetric mode of
EFOSC2 is lacking the full characterisa-
tion needed to reach ultra-high and abso-
lute precision polarimetric measure-
ments under the conditions of our experi-
ment. Using special calibration tech-
niques, in some astronomical observa-
tions it is already possible to achieve a
precision of 10
6
(e.g., Bailey et al., 2010),
but in our experiments we are limited
by an instrument setup in which most of
the signal comes from off-axis rays. Patat
& Romaniello (2006) and Bagnulo et al.
(2009) have discussed the spurious polar-
imetric effects observed off-axis with
the FORS instrument of the VLT. Some
of these effects may also play a role
in EFOSC2. We must also remark that
the prepared samples are not fully repro-
ducible in their confguration, and thus
likely introduce some effects due to their
different backscattering properties.
In conclusion, the polarised spectra of
Figures2 and 3 show that the polari-
metric techniques employed by us are
sensitive to the presence of biotic mate-
rial. We detect chiral signatures of pig-
ments involved in biotic photosynthetic
reaction chains. Signatures of chlorophyll
pigments can be captured both in sam-
ples of leaves and cyanobacteria of spe-
cies Chroococcidiopsis. More accurate
measurements of the polarised signal
amplitudes require experiments to be
carried out with samples prepared under
better controlled laboratory conditions.
It will also be interesting to observe the
polarimetric signatures of Chroococcidi-
opsis in their natural habitat beneath
translucent rock surfaces. This will allow
the feasibility of detecting this type of
bacteria on other Solar System bodies in
the future with extremely large telescopes
to be assessed.
We thank all the La Silla staff who sup-
ported the setup and execution of this
unusual experiment! Stefano Bagnulo
acknowledges a grant received as visiting
scientist at ESO/Chile.
References
Bagnulo, S. et al. 2009, PASP, 121, 993
Bailey, J. et al. 1998, Science, 281, 5377, 672
Bailey, J., Lucas, P. W. & Hough, J. H. 2010,
MNRAS, 405, 2570
Gregory, R. P. F. & Raps, S. 1974, Biochem. J.,
142, 193
Friedmann, E. I. & Ocampo-Friedmann, R. 1995,
Adv. Space Res., 15(3), 243
Patat, F. & Romaniello, M. 2006, PASP, 118, 146
Saviane, I. et al. 2007, The Messenger, 129, 14
Sparks, W. B. et al. 2009, PNAS, 106 (19), 7816
Wolstencroft, R. D., Tranter, G. E. & Le Peleven, D. D.
2004, Bioastronomy 2002, IAU Symp., 213, 149
Links
1
http://www.astro.puc.cl/astrobio2010
Figure 2. The Philoden-
dron leaf is shown at left
and its polarised spec-
trum measured with
EFOSC2 at middle and
right. The blue lines
show the circular polari-
sation (V/I). In the right-
most panel, V/I is recti-
fed to zero. The green
lines show the null pro-
fle (offset by 0.06%
in the rightmost panel,
for display purposes)
demonstrating the sta-
tistical errors. The black
line is the refectivity
(in arbtirary units).
Figure 3. The sample of
Chroococcidiopsis is
shown at left and its
polarised spectrum in
the middle and right
panels. The blue line
shows circular polarisa-
tion and the green line
is the null profle (offset
by +0.02%). The red
line shows the spectrum
obtained with the re-
tarder waveplate at
position angles offset by
45, rectifed to zero. In
the rightmost panel, V/I
is rectifed to zero.
450
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
500 550 600
Wave|ength (nm}
v
/
l
N
v

(
%
}
650 700 750 650 700 750
Wave|ength (nm}
-0.1
0.1
-0.05
0
0.05
-0.04
-0.02
0
0.02
450 500 550 600 650 700 750
0.04
v
/
l
N
v

(
%
}
Wave|ength (nm}
650 700
Wave|ength (nm}
-0.01
0.02
0
0.01
28
This abundance pattern between ONa
and AlMg is characteristic of GCs,
while it is almost absent among feld stars
(Gratton et al., 2000). It is not limited
to red giants, but is present also in the
low-mass MS stars, as demonstrated
by the key study with the VLTs high reso-
lution optical spectrograph, UVES, by
Gratton et al. (2001; based on the large
programme 165.L-0263). These un -
evolved stars cannot have sustained the
nucleosynthesis chains that deplete O
and Mg, and enhance Na and Al, because
they cannot reach the required tempe-
ratures and have very thin convective en -
velopes, unable to mix nuclear prod-
ucts into their atmospheres. The UVES
large programme excluded early claims
that this composition might result from a
peculiar evolution of GC stars: all low
O/high Na stars originated from matter
processed and ejected by stars belong-
ing to a previous stellar generation within
the GC, although not by the SNe (or
the low-mass stars, which give a different
characteristic imprinting) of this earlier
population.
Self-pollution models still lag behind the
observations and are not yet able to
explain convincingly all observed features:
candidate frst generation (FG) polluters
Raffaele Gratton
1
Eugenio Carretta
2
Angela Bragaglia
2
Sara Lucatello
1
Valentina DOrazi
1
1
INAFOsservatorio Astronomico di
Padova, Italy
2
Bologna, Italy
In the last few years, it has become evi-
dent that globular clusters, previously
accepted as prime examples of simple
stellar populations, contain at least
two, and in some cases more, stellar
generations. Thanks to the superb spa-
tial resolution of the Hubble Space
Telescope on one side, and to the fore-
front spectroscopic and multiplexing
capabilities of UVES and FLAMES at the
VLT on the other, extensive, high preci-
sion datasets are shedding light on clus-
ter formation and evolution. Here, we
briefy describe the contribution by our
team to these exciting discoveries.
Multiple populations in globular clusters
Globular clusters (GCs) are very massive
stellar complexes. As an example, Fig-
ure1 shows an image of the globular clus-
ter 47Tucanae. All GCs observable in
or near the Milky Way are very old, so their
formation cannot be observed in detail.
These early phases were likely complex,
including a variety of dramatic and ener-
getic phenomena such as supernova
(SN) explosions, photoionisation, high and
low velocity winds from blue and red
massive stars, possibly combining in giant
expanding bubbles of gas and shock
fronts that may have triggered further star
formation. The scene where this drama
occurred might have been even more
di versifed in the core of giant clouds
or, for the most massive clusters, even
of dwarf galaxies. These dramatic events
left a trace represented by the chemi-
cal composition of the stars that not
only can be followed today in quite subtle
spectral features, but also can infuence
their distribution along the main se quence
(MS) and the horizontal branch (HB).
All GCs investigated to date show a pecu-
liar abundance pattern (for a review, see
Gratton et al., 2004). They usually appear
to be very homogeneous, with a few
exceptions like M54 and Omega Centauri,
insofar as the Fe-peak elements are con-
cerned, thus ruling out self-enrichment by
SN ejecta. However, many GC stars have
an unusual composition, rich in Na and
Al and poor in O and Mg, typical of mate-
rial that experienced H-burning at very
high temperatures, while others have a
normal O- and Mg-rich but Na- and Al-
poor composition, which is virtually iden-
tical to that of metal-poor feld stars.
Within each GC, the abundances of Na
and O, or Al and Mg are anti-correlated
with each other (see Figure2). Since
the main outcome of H-burning is He, we
expect low O/high Na stars also to be
He-enriched, which explains the occur-
rence of multiple HBs and MSs. The
connection between ONa and MgAl
anti-correlations, variations in He abun-
dances, and the presence of multiple
HBs is however likely to be more complex
than described here (see Gratton et al.,
2010).
Observations of Multiple Stellar Populations in
Globular Clusters with FLAMES at the VLT
Figure 1. Image of the globular cluster 47Tuc
(NGC104), obtained with the ESO 1-metre Schmidt
Telescope at the La Silla Observatory in Chile.
29
are either thermally pulsating intermediate-
mass asymptotic giant branch (AGB)
stars undergoing hot bottom burning
(Ventura et al., 2001) or massive rotating
stars prior to SN explosion (Decressin et
al., 2007).
Our FLAMES survey
Although the star-to-star ONa abun-
dance anticorrelation had been recog-
nised early on as a pivotal signal of
self-enrichment in GCs, signifcant pro-
gress was hindered by the slow acqui-
sition rate of data, due to the use of
single object spectrographs. At the time
of the Gratton et al. (2004) review, data
for a grand total of only about 200 stars,
distributed in some ten GCs, were availa-
ble. The high multiplexing capability of
the VLTs Fibre Large Array Multi Element
Spectrograph (FLAMES) allowed us
an order of magnitude increase in data
collection capability: our survey has
already harvested spectra for more than
2000 red giants, analysed individually
with a homogeneous procedure. With our
work, we intended to answer some fun-
damental questions: Were GC stars really
born in a single instantaneous burst?
Did all GCs self-enrich themselves? How
do abundance patterns within each
individual GC relate to the formation and
early evolution of the GC itself and of each
individual member?
We selected a large sample of GCs with
diverse HB morphology, since, as men-
tioned, there appears to be a link between
the chemical anomalies, the extension
of the HB, and the presence of multiple
MSs. We obtained FLAMES data on
24 GCs (ESO programmes 072.D-0507,
073.D-0211, 081.D-0286, and 083.D-208).
We used the pipeline-reduced spectra
to obtain Fe, Na, and O abundances from
spectra from the medium-high resolution
spectrograph, GIRAFFE, and of more
elements, including Mg and Al, from the
UVES data. Up to now we have published
about 15 refereed papers presenting re -
sults for 21 GCs.
With this data, we have confrmed that
all GCs contain multiple populations
(Carretta et al., 2009a). We even pro-
posed a new defnition of bona fde GCs
as stellar aggregates showing the NaO
anticorrelation, as distinct from associa-
tions and open clusters (Carretta et al.,
2010a). In Figure3, adapted from Carretta
et al. (2010a), we show massive stellar
clusters (both GCs and open clusters) in
the relative age vs. mass plane. Here
relative ages are used (see Carretta et al.,
2010a): in this scale, a relative age of
1 means an age close to 12.5 Gyr. Differ-
ent symbols are used for open clusters
and GCs, and for those clusters where
the NaO anticorrelation has been found
or not (for several clusters, data currently
available are not suffcient to clarify this
point). From this fgure, it seems clear
that the presence of multiple populations
is the typical result of the formation pro-
cess of massive stellar clusters and must
be explained by their formation scenarios.
The size of our sample led, for the frst
time, to a quantitative estimate of the
fraction of FG and second generation
(SG) stars in GCs (Carretta et al., 2009a).
The SG is always dominant, including at
least two thirds of the total population.
Since the peculiar chemical compo-
sition of the SG stars indicates that only
a fraction of FG stars may have pro-
duced the right gas mixture, the original
clouds from which the GCs formed
should have been much more massive
than current GCs (likely ten times or even
more). Since GCs now account for
roughly 1% of the Galactic halo, a large
fraction of the halo should have origi-
nated in the same episodes that led to
the formation of the GCs.
We obtained diverse distributions of the
NaO and MgAl anticorrelations for
different GCs (Carretta et al., 2009a, Fig-
ure2; Carretta et al., 2009b) in some
of the GCs, indicating that some parame-
ter, most likely the typical polluter mass,
is varying from cluster to cluster. There is
a strong link between the extension of
the NaO anticorrelation and some of the
main parameters of GCs, mainly mass
and metallicity.
We also confrmed that in spite of these
striking star-to-star differences in the
light elements, most GCs (with a few
notable exceptions) are extraordinarily
homogeneous in Fe-peak elements, with
upper limits as low as 5% on the root
0 1
1
-1
-0.5
0.5
0
(0/Fe}
NGO 6809
(
N
a
/
F
e
}
0 1 -1
(0/Fe}
NGO 6838
0 1 -1
(0/Fe}
NGO 7078
0 1 -1
(0/Fe}
NGO 7099
0 1 -1
(0/Fe}
NGO 1851
1
-0.5
0.5
0
NGO 6254
(
N
a
/
F
e
}
NGO 6388 NGO 6397 NGO 6441 NGO 6752
1
-0.5
0.5
0
NGO 4590
(
N
a
/
F
e
}
NGO 5904 NGO 6121 NGO 6171 NGO 6218
1
-0.5
0.5
0
NGO 104
(
N
a
/
F
e
}
NGO 288 NGO 1904 NGO 2808 NGO 3201
Figure 2. The NaO anticorrelation is shown as
log(Na/Fe) vs. log(O/Fe) normalised to the solar
abundance in 20GCs as observed with FLAMES
(adapted from Carretta et al. [2009a] and Carretta
et al. [2010]). The typical error bar for each set of
measurements is shown.
30
HB, or the blue and red sequences of
GCs seen in many galaxies), and the
relation between GCs and the nuclei of
dwarf galaxies, and many more.
With its large set of very competitive in -
struments, the VLT is likely to play a
fundamental role in this game even in the
future. UVES and FLAMES observa-
tions were essential in revolutionising our
view of the GCs. New perspectives are
possible with the high sensitivity of
X-shooter, revealing the spectra of faint
MS stars (see Bragaglia et al., 2010b), or
the access to near-infrared spectra pro-
vided by CRIRES.
References
Bellazzini, A. et al. 2008, MSAIt, 79, 663
Bragaglia, A. et al. 2010a, A&A, 519, 60
Bragaglia, A. et al. 2010b, ApJL, 720, L41
Carretta, E. et al. 2009a, A&A, 505, 117
Carretta, E. et al. 2009b, A&A, 505, 139
Carretta, E. et al. 2009c, A&A, 508, 695
Carretta, E. et al. 2010a, A&A, 516, 55
Carretta, E. et al. 2010b, ApJL, 714, L7
Decressin, T. et al. 2007, A&A, 464, 1029
DOrazi, V. & Marino, A. F. 2010, ApJL, 716, L166
DOrazi, V. et al. 2010, ApJL, 719, L213
Gratton, R. et al. 2000, A&A, 354, 169
Gratton, R. et al. 2001, A&A, 369, 87
Gratton, R. et al. 2004, ARA&A, 42, 385
Gratton, R. et al. 2010, A&A, 517, 81
Lata, S. et al. 2002, A&A, 388, 158
Ventura, P. et al. 2001, ApJ, 550, L65
mean square dispersion. A by-product
of our work is a new metallicity scale for
GCs based on homogeneous abun-
dances from high resolution UVES spec-
tra (Carretta et al., 2009c). The homo-
geneity and good statistics also allowed
the frst distinction between the He-rich,
SG and the He-poor FG populations
at the red giant branch (RGB)-bump
(Bragaglia et al., 2010a) to be made, as
foreseen by models.
Our frst extensive study of the chemical
composition of M54 in the nucleus of
the disrupting Sagittarius dwarf galaxy
showed a metallicity dispersion and NaO
anticorrelation in both the metal-rich
and metal-poor component (Carretta et
al., 2010b), with similarities with Omega
Centauri, probably indicative of a similar
origin in dwarf spheroidal galaxies.
The Li abundances offered a comple-
mentary approach. Li is easily destroyed
in stellar interiors so, if there is no Li
production within the polluters, Li and O
should be positively correlated and Li
and Na anticorrelated. Measures of O, Na,
and Li abundances in the same stars
are rare, but DOrazi and Marino (2010)
found no LiO correlation in a sample
of about 100 giants in M4. If confrmed in
other GCs, this would be explained
only with Li production in the polluters,
very likely intermediate-mass AGB stars.
Finally, we found that the fraction of
Ba-stars (which arise from mass transfer
in binaries) is higher among FG stars
(DOrazi et al., 2010). This prompted us to
estimate the binary fraction from multi-
epoch measurements of radial velocities,
available for three of our GCs, fnding that
the binary fraction of SG stars is much
lower than that of FG stars. Binaries are
more common in low than in high density
environments: our fnding is then a direct
probe of the ambient condition at the dis-
tant epochs where the bulk of different
stellar generations formed in GCs.
These results are unprecedented and
were made possible by the large samples
of high resolution spectra available,
and by the high degree of homogeneity
attained by the analysis procedures
adopted in our survey.
Future directions
Our results, as well as the discovery of
multiple MSs from high precision
Hubble Space Telescope photometry, are
opening unexpected and fascinating
windows on the quest for the formation
and evolution of massive stellar clus-
ters. However, several issues remain
presently unsolved, like the nature of the
polluters, the overall timescale of the
self-enrichment, the existence of material
with pristine composition diluting the
products of the polluters, the relation
between the formation of GCs and that of
the Galactic Halo, the role of different
populations on the dynamical evolution of
GCs, the connection between multiple
populations and other properties of GCs
(e.g., mass, the second parameter on the
Figure 3. The relative age parameter is plotted vs.
absolute magnitude M
V
for globular and old open
clusters. Red flled pentagons are for GCs where the
NaO anticorrelation has been observed. Open stars
mark GCs for which insuffcient data about the NaO
anticorrelation is available. Squares are for the clus-
ters of the Sagittarius dSph galaxy and open circles
are for old open clusters (see Carretta et al., 2010a).
Green crosses mark those clusters where the NaO
anticorrelation has been searched for but not found.
Superimposed are lines of constant mass (light solid
lines, see Bellazzini et al., 2008). The heavy blue
solid line (at a mass of 4 10
4
M
A
) is the proposed
separation between globular and open clusters.
Re|at|ve age
0.6
0.6
-2
-4
-6
-8
-10
2 10
6

2 10
5

2 10
4

2 10
3
0.8 1
Gratton R. et al., Observations of Multiple Stellar Populations in Globular Clusters
31
Simon Clark
1
Ignacio Negueruela
2
Ben Ritchie
1
Paul Crowther
3
Sean Dougherty
4
1
Department of Physics and Astronomy,
The Open University, Milton Keynes,
United Kingdom
2
Departamento de Fsica & Ingeniera,
Universidad de Alicante, Spain
3
University of Sheffeld, United Kingdom
4
National Research Council, Herzberg
Institute for Astrophysics, Penticton,
Canada
Westerlund1 is the frst example of
a super star cluster identifed within the
Galaxy. As such, its proximity allows
us to resolve directly and study individ-
ual stars down to sub-solar masses,
as well as their complex interactions.
This in turn permits advances in our
un derstanding of the physics of these
stellar powerhouses, which drive evo-
lution in starburst galaxies near and far.
Here we provide a brief overview of
our current understanding of this clus-
ter, both in terms of its stellar constit-
uents and the constraints they place
on the evolution of massive stars from
cradle to grave and its global proper-
ties.
Starburst galaxies and super star clusters
Images of starburst galaxies such as
the Antennae, M82 and NGC1313 (see
Figure1) reveal that the process of star
formation appears hierarchical, with stars
forming within massive clusters, which
in turn are located within larger complexes
that refect the underlying structure of
the natal giant molecular clouds. The im -
pact of both these super star clusters
(SSCs) and complexes on the wider galac-
tic ecology is immense: the OB stellar
population contained within dominates
both the ultraviolet (UV) and, via dust re -
processing, the infrared (IR) radiative
output of the host galaxy, while their post-
supernovae (post-SNe) relativistic rem-
nants neutron stars and black holes
within binary systems are responsible
for the high energy emission.
Moreover, the same population of mas-
sive stars provides signifcant input
of mechanical energy and chemically
processed material via their stellar winds
and SNe. Indeed, the combination of
both radiative and kinetic feedback is
thought to be responsible for the initiation
of superwinds that may promote or sup-
press subsequent generations of star
formation, as well as enrich the inter-
galactic medium. Nevertheless, despite
their pivotal role in galactic evolution the
small physical extent of SSCs in exter-
nal galaxies means that they must be
studied via comparison of their integrated
spectral and photometric properties.
But such an approach is based on two
unproven hypotheses: (i) that the Initial
Mass Function (IMF) of stars within such
clusters (and complexes) is identical to
that determined locally; and (ii) that stars
within such clusters follow comparable
evolutionary paths to those within
the Local Group, such that we under-
stand their radiative output, lifetimes and
post-SNe endpoints. Without such
assumptions we cannot accurately cali-
brate the population synthesis codes
used to determine cluster ages and inte-
grated masses (and hence star formation
rates), nor constrain the degree of
mechanical and radiative feedback that
yields galactic-scale superwinds.
Unfortunately, through the 20th century
it had been supposed that the Gal-
axy lacked spatially-resolved examples of
SSCs, with the integrated masses of
young (<20 Myr) massive open clusters
typically being less than 10
3
M
A
com-
pared to 10
5
10
7
M
A
for the former. Tar-
geted near-IR observations of the centre
of the Galaxy had revealed the presence
Dissecting the Galactic Super Star Cluster
Westerlund1 A Laboratory for Stellar Evolution
Figure 1. The central region of the vigorous starburst
galaxy NGC1313 is shown in this colour image
from VLT FORS data (from Release eso0643). Note
the numerous shells of ionised gas excited by young
massive clusters similar to Wd1.
32
the presence of low excitation metallic
and molecular features for cool objects,
was constructed from both real and
synthetic spectra (Clark et al., 2005;
Negueruela et al., 2010). As well as con-
frming the classifcations of Westerlund
(1987), these observations also identi-
fed large additional populations of OB
supergiants and WolfRayet stars (WRs),
which had previously escaped detection
due to a combination of comparatively
low optical luminosity and heavy redden-
ing. Indeed, subsequent observations
revealed a population of 25WRs, com-
posed of both WN and WC subtypes; the
richest haul detected within a Galactic
cluster at this time (Crowther et al., 2006).
These observations enabled us to deter-
mine that Wd1 has an age of ~45Myr
and appears to be co-eval; conclusions
confrmed by later observations (e.g.,
Negueruela et al., 2010). Moreover, utilis-
ing these spectra to calibrate our pho-
tometry, we were able to estimate that
>>100 stars within the cluster must
have evolved from progenitors with initial
of the Arches, Quintuplet and Galactic
Centre clusters with masses ~10
4
M
A
;
still an order of magnitude smaller than
known SSCs. In fact the frst known
example of a SSC had been identifed
fully 40 years earlier indeed before
their widespread recognition in external
starburst galaxies! but had subse-
quently remained in relative obscurity.
Westerlund1
Located within the constellation of Ara
(the Altar) Westerlund1 (Wd1; see
Figure2) was simply described by Bengt
Westerlund in 1961 as a very young,
heavily reddened cluster. Indeed the
high extinction towards Wd1 (A
v
~11 mag)
hampered spectroscopic investigation,
with the frst such survey following over a
quarter of a century later (Westerlund,
1987). Despite revealing an unprece-
dented population of high luminosity
supergiants of both early and late spec-
tral types, Wd1 once again sank back
into obscurity until radio observations of
the B[e] supergiant Wd1-9 serendipi-
tously detected a large number of radio
sources amongst the evolved stellar
population (Figure3; see Clark et al.,
1998 and Dougherty et al., 2010). Unex-
pectedly, a number of the cool red super-
giants (RSGs) and yellow hypergiants
(YHGs) were found to be strong radio
sources, despite lacking the requisite UV
fux to ionise their ejecta.
Prompted by these results we undertook
spectroscopic and photometric ob -
servations of cluster members between
20012; frstly with the ESO 1.52-metre
telescope and the Boller & Chivens spec-
trograph, and subsequently at higher
signal-to-noise ratio (S/N) and resolution
with the New Technology Telescope
and EMMI spectrograph (Clark et al.,
2005). Given the reddening towards Wd1,
classifcation spectra were obtained
from 600900 nm, rather than the more
common 400600 nm window. An
appropriate classifcation scheme, based
on the occurrence and strength of HI,
He I and He II lines in the OBA stars and
Figure 2. Optical image of Wd1 obtained with
the Wide Field Imager mounted on the MPG/ESO
2.2-metre telescope. Note the high degree of
reddening to the cluster in comparison to the fore-
ground Bsupergiant (lower right). Image from
Release eso1034c.
16
h
47
m
15
s
-52o
265
-51o
4550o
D
e
c
|
|
n
a
t
|
o
n

(
J
2
0
0
0
}
R|ght Ascens|on (J2000}
47
m
10
s
47
m
05
s
47
m
00
s
75
5
15
16b
72
16a
17
70b
26
31b
9
32
44
20
239
243
237
19b
12a
4
Figure 3. Overplot of radio observations (contours)
on an optical image of Wd1. Note the emission
associated with the cool hypergiants (Wd1-4, 12a,
16a, 32 & 265 YHGs; Wd1-20, 26 & 237 RSGs)
indicative of signifcant ongoing mass loss.
Clark S. et al., Dissecting the Galactic Super Star Cluster Westerlund1
33
masses 30 M
A
. Such a census has
two implications. Firstly, it offers an
explanation for the anomalous radio emis-
sion associated with the YHGs and
RSGs, whereby the diffuse UV radiation
feld from the hot stellar population
(OB stars and WRs) provides the ionising
photons. Secondly, using this popula-
tion to normalise a Kroupa-type IMF we
were able to infer a total (initial) mass for
Wd1 of ~10
5
M
A
, making it the most
massive Galactic cluster yet discovered
by an order of mag nitude and hence the
frst SSC within the Milky Way.
Therefore, Wd1 is the frst example of an
SSC for which its relative proximity per-
mits the resolution and study of individual
stars to sub-solar masses enabling
the ecology of such an agglomeration to
be decoded for the frst time.
Massive stellar evolution a tale of two
pathways
A key driver in massive stellar evolution
is mass loss, which strips away the H-rich
mantle of O stars to yield H-depleted
WRs. Indeed the rate at which this pro-
ceeds not only governs the precise evo-
lutionary path trodden by the star, but
also its ultimate post-SNe fate. Radio ob -
servations had already revealed the
characteristic signature of heavy mass
loss associated with many stars (Clark et
al., 1998; Dougherty et al., 2010), while
our new spectroscopy reveals a rich post-
main sequence (post-MS) population.
Thus Wd 1 presents a unique opportunity
to investigate the physics of stellar evo-
lution for some of the largest stars to be
found within the Galaxy.
Nevertheless, such a goal is challenging
for three reasons. Firstly, sophisticated
modelling with non-local thermodynamic
equilibrium (non-LTE) atmospheric codes
is required to determine accurately physi-
cal properties such as stellar luminosity,
temperature, mass-loss rate and chemi-
cal abundances; all of which are essential
to determine accurately the precise evo-
lutionary state of an individual star (see
Figure4; Ritchie et al., 2009b). Secondly,
as such stars transit from the MS to
H-depleted WR phase they pass through
a dizzying variety of short-lived evolu-
tionary states that include the luminous
blue variables (LBVs) and the closely
related PCygni supergiant/Bhypergiant,
supergiant B[e] stars, YHG and RSG
stages. It is thought that mass loss in
such phases may play a dominant role in
the removal of hydrogen and subse-
quent formation of WRs, but it appears
likely that this process is accomplished
via transient outbursts in which mass-
loss rates increase by several orders of
magnitude over their quiescent values.
A prime example of this behaviour in
LBVs is the 19th-century eruption of
Car, while similar behaviour has also
been inferred for YHGs and RSGs such
as Cas and VY CMa. However, such
outbursts are rare, with none subject to
modern day observational and analytical
techniques, meaning that the physical
mechanism leading to the extreme mass-
loss events has yet to be identifed.
Finally, a synthesis of our multi-wavelength
observations of Wd1 suggests a very
high binary fraction amongst the WR pop-
ulation (Crowther et al., 2006; Clark
et al., 2008). Binary-mediated mass loss
through Roche-lobe overfow in short
period systems offers an additional evolu-
tionary pathway for massive stars,
which leads to signifcantly lower pre-SNe
core masses than expected for single
stars. The importance of this pathway was
dramatically demonstrated by the dis-
covery of a magnetar (a class of neutron
star with magnetic feld strengths
~10
15
G) within Wd1 (Muno et al., 2006).
Conventional wisdom assumes that stars
with initial masses in excess of ~25M
A

give rise to the post-SNe formation of
black holes rather than neutron stars.
Consequently, with a current MS turnoff
mass in excess of ~30M
A
(Clark et
al., 2005; Negueruela et al., 2010), it had
been expected that only black holes
should currently be forming within Wd1.
Indeed, given that the pre-SNe mass
loss required to yield a neutron star rather
than a black hole is greatly in excess
of that expected for single star evolution,
signifcant binary driven mass loss
appears mandatory to allow for the for-
mation of the Wd1 magnetar.
Quantitative analysis via VLT follow-up
The existing spectral data were insuff-
cient to address these issues, being of
too low S/N and resolution to enable
non-LTE-model atmosphere analysis and
only at a single epoch, thus preventing
the identifcation of either intrinsic (insta-
bilities) or extrinsic (binary) variability.
Consequently, throughout the period
20049 we obtained multiple epochs of
high quality, 600900 nm spectroscopic
data of ~100 members of the evolved
stellar population of Wd1 with both FORS
and FLAMES on the VLT. Full details of
the target selection, experimental setup
and reduction techniques employed may
be found in Ritchie et al. (2009a) and
Negueruela et al. (2010). These data ena-
bled us to address each of the challenges
described above.
867 868 869 870
Wave|ength (nm}
777 777.25 777.5 777.75
Wave|ength (nm}
N
o
r
m
a
|
|
s
e
d

f
u
x
871 872
Nl
DlB
Ol
0.5
0.4
0.8
0.7
0.6
1
0.9
1.1
0.5
0.4
0.8
0.7
0.6
1
0.9
1.1 Figure 4. Comparison of
observed (black) and
synthetic (red) spectra of
the LBV Wd1-243,
focusing on the NI and
OI triplets and indicat-
ing signifcant chemical
evolution of N and O
(respectively 12 and 0.1
times solar abundances).
34
(FLAMES) the identifcation of radial
velocity variables resulting from binary
motion. While analysis of the full dataset
is currently underway, preliminary re-
sults for the frst year of observations of
the brightest quarter of targets imply a
binary frequency of >40% amongst the
OB-supergiant population (Ritchie et al.,
2009a). This fnding suggests that single
and binary channels may be of compa-
rable importance in the evolution of mas-
sive stars, although a determination of
the period distribution of the binary popu-
lation will be necessary to quantitatively
confrm this assertion.
A second critical result from this analysis
was the identifcation of Wd1-13 which
had previously been identifed as an
Firstly, the data permit quantitative atmos-
pheric modelling of individual stars
in order to determine their stellar parame-
ters. An example of such an analysis
is presented in Ritchie et al. (2009b; see
Figure4). By constraining the chemical
abundances for the LBV Wd1-243 it
was possible to demonstrate unambigu-
ously that it is a highly evolved, likely
post-RSG, object. The use of elemental
abundances to place the panoply of
post-MS stars LBVs, B/YHGs, RSGs,
etc. in a precise evolutionary scheme
is a powerful technique, with mass loss
systematically driving down the H/He
ratio while simultaneously exposing the
products of nuclear burning at the stel-
lar surface. We aim to extend this ap -
proach to the remaining population of
transitional stars within Wd1 in the imme-
diate future.
Secondly, in combination with existing
historical observations, the data ena-
bled a search for variability amongst the
subset of bright evolved stars over a
~50-yr baseline. Such an approach had
previously allowed for the classifcation
of Wd1-243 as an LBV, while radio obser-
vations of the B[e] star Wd1-9 showed
an episode of enhanced mass loss that
likely ended within the last 200 years
(Dougherty et al., 2010). Intriguingly, while
signifcant wind variability and pulsa-
tional instability appeared to be ubiqui-
tous for all subtypes of evolved stars
observed within Wd1, no further exam-
ples of LBVs have yet been identifed
within the hot super-/hyper-giant popula-
tion.
However dramatic variations in spectral
type, likely refecting pulsation-driven
changes in the stellar photosphere, were
identifed amongst the Blue-/YHG and
RSG populations (c.f. the YHG Wd1-265
shown in Figure5; Clark et al., 2010);
indeed all stars of spectral types later
than B1 appear to be pulsationally unsta-
ble. Amongst the cool super-/hyper-
giants this occurrence is of considerable
importance since identical behaviour
has been observed in feld YHGs such as
Cas, where it has been associated
with episodes of dramatically enhanced
mass loss (~5 10
2
M
A
yr
1
; Lobel et
al., 2003). When coupled with quantitative
modelling, continued spectral monitoring
of these stars raises the prospect of
both determining the duty cycle of such
mass-loss events as well as providing
stringent constraints on the underlying
physics driving these instabilities. Like-
wise, high cadence observations of
the hotter supergiants potentially open up
their internal structure to scrutiny via
asteroseismology studies, while also prob-
ing the origin of their highly structured
stellar winds.
Finally, we turn to the main science
driver for our intensive spectrographic
survey comprising multi-epoch obser-
vations of ~100 evolved stars over a
14-month baseline utilising the VLT Fibre
Large Array Multi Element Spectrograph
Figure 5. Time-resolved
spectra of the pulsating
YHG Wd1-265. Compar-
ison with classifcation
spectra (dotted lines)
indicates signifcant vari-
ability in spectral type
over a period of only 46
days.
864 866 868 870
Wave|ength
N
o
r
m
a
|
|
s
e
d

f
u
x

(
+

o
f
f
s
e
t
}
872 874 876
29/06/2008
14/08/2008
Pa12
S| l S| l
A9 |a
F2 |a
F5 |ab
Mgl Mgl Fel Fel Sl
Pa13
N l
0.5
0
1
1.5
2
Figure 6. Radial velocity
curve for both compo-
nents of the WNL+OB
binary Wd1-13 folded on
the ~9.27-day orbital
period.
-1
0
100
200
-0.2
-300
-200
0 0.2 0.4 0.6 0.8 1 1.2
Phase
Absorpt|on
Em|ss|on
R
a
d
|
a
|

v
e
|
o
c
|
t
y

(
k
m
/
s
}
Clark S. et al., Dissecting the Galactic Super Star Cluster Westerlund1
35
eclipsing system as a double-lined
WNL+OB supergiant binary (Figure6).
This combination of properties enabled
us to determine dynamical masses of
23 3 and 35 5 M
A
for the WNL and
OB supergiant components respectively
(Ritchie et al., 2010). Comparison to theo-
retical models of WR binary evolution
suggest that the WNL component had an
initial mass of ~40M
A
, which immedi-
ately places a frm lower limit on the mass
of the magnetar progenitor. Given that
magnetars are also present in clusters
with comparatively low-mass MS turnoffs
(<20M
A
; Clark et al., 2008), their pro-
genitors clearly span a range of masses,
implying that an additional ingredient
such as rapid rotation or a high natal
magnetic feld must be required for their
formation.
Moreover, we may also place quantitative
constraints on the location of the bifurca-
tion in the canonical Conti scenario for
stellar evolution, whereby the most mas-
sive stars evolve via:
O MS WNLh LBV WNL WNE
WC SNe

thus avoiding a cool hypergiant phase,
and less massive stars via
O MS OB SG RSG LBV/YHG?
WNL WNE(WC?) SNe
(where the WNLh designation means that
the star shows hydrogen emission lines
in its spectrum). Given the both YHGs and
RSGs are currently present within Wd1,
the masses implied for the progenitors of
these stars by Wd1-13 reveals that the
division must occur at >40M
A
.
Future prospects
In the last decade of study Wd1 has
yielded many of its secrets, enabling us
to confrm that it is the frst SSC to be
identifed within the Galaxy, as well as
permitting us to place powerful observa-
tional constraints on the evolution of
massive stars in their natural environment
for the frst time. However, what does
the future hold? Our immediate goal is to
fully determine the physical properties of
the binary population of Wd1 in terms
of frequency, orbital separation and mass
ratio, which will be accomplished by
comparison of the complete RV dataset
to Monte-Carlo simulations. It is hard
to overestimate the importance of such a
goal in addition to constraining the
relative weighting of single and binary evo-
lutionary channels accurately, such infor-
mation will also constrain the physics
of massive star formation (e.g., Ritchie et
al. [2009a] and references therein) as
well as the production effciency of both
high- and low-mass X-ray binaries, bi-
nary pulsars a major source of gravita-
tional radiation and potentially the pro-
duction of gamma-ray bursts.
Looking further ahead, the presence of
RSGs within Wd1 permits a clear and un -
equivocal test of current theories of mas-
sive stellar evolution, which do not permit
such stars to exceed log(L/L
A
)5.6
(e.g., Meynet & Maeder, 2005). Unlike
feld stars, the well-determined distance
to Wd1 permits an accurate determina-
tion of the luminosity of the cluster RSGs
once an appropriate bolometric cor-
rection has been determined via model
atmosphere analysis.
Turning to Wd1 as a whole, we have a
unique opportunity to investigate the in -
teraction between individual massive
stars within an SSC. The motivation for
such a study is vividly illustrated by the
cometary nebulae associated with the
RSGs Wd1-20 & 26 and which are visible
in mid-IR and radio images (Figure3;
Dougherty et al., 2010). These appear to
be the result of the ablation and en -
trainment of the outer stellar layers/winds
of these stars by the incipient cluster
wind driven by the mechanical and radia-
tive feedback from individual stars
and SNe. But how do such winds work?
Comparison of radio, mid-IR and X-ray
data indicate that the intercluster medium
appears to be multi-phase, composed
of cool, neutral and dust-laden clumps
shadowing warmer ionised gas in close
proximity to the core, which, in turn,
are imbedded in an X-ray-bright compo-
nent, emitting via both thermal and non-
thermal mechanisms. How is momen-
tum imparted to this material to yield a
cluster wind and what is the effciency
of this process? Answers to both ques-
tions are essential if we are to understand
the impact of SSCs on their wider (extra-)
galactic environments.
And fnally regarding Wd1 in a wider con-
text did it form alone? Examination
of star-forming regions suggests that such
clusters do not form in isolation, but
currently there is no evidence for ongoing
star formation closely associated with
Wd1. Was it born with other siblings which
have since dispersed, or did it instead
form monolithically in a single starburst
event and if so, why? How many other
examples lurk in the Galactic plane?
Systematic spectroscopic follow-up of
candidates identifed via current and
future surveys such as the VISTA Varia-
bles in the Via Lactea survey will enable
us to place and understand Wd1 in the
context of the recent star formation his-
tory of the Milky Way. Clearly the discov-
ery and analysis of Westerlund1 has
answered many questions, but has raised
many more.
References
Clark, J. et al. 1998, MNRAS, 299, L43
Clark, J. et al. 2005, A&A, 434, 949
Clark, J. et al. 2008, A&A, 477, 147
Clark, J. et al. 2010, A&A, 514, 87
Crowther, P. et al. 2006, MNRAS, 372, 1407
Dougherty, S. et al. 2010, A&A, 509, 79
Lobel, A. et al. 2003, ApJ, 583, 923
Meynet, G. & Maeder, M. 2005, A&A, 429, 581
Muno, M. et al. 2006, ApJ, 636, L41
Negueruela, I., Clark, J. & Ritchie, B. 2010, A&A,
516, A78
Ritchie, B. et al. 2009a, A&A, 507, 1585
Ritchie, B. et al. 2009b, A&A, 507, 1597
Ritchie, B. et al. 2010, A&A, 520, 48
Westerlund, B. 1987, A&AS, 70, 311
36
mental relation between metallicity, mass
and SFR observed in local galaxies. This
fnding suggests that the same mecha-
nisms of galaxy formation are at work
at any epoch in the redshift interval
0<z<2.5.

Until recently, little was known about the
metallicity of galaxies at z>3, due to
the diffculty of detecting the optical neb-
ular lines (redshifted into the near-infra-
red) required to measure the metallicity in
these faint systems. However, this is
a crucial epoch of very fast galaxy evolu-
tion, just before the peak of cosmic star
formation, which requires thorough inves-
tigation to understand the formation of
primeval galaxies properly.
The AMAZE and LSD surveys
We have undertaken two major projects
using the near-infrared integral feld
spectrograph SINFONI at the VLT. AMAZE
(Assessing the Mass-Abundances
redshift [Z] Evolution) is an ESO large
programme that was awarded 180 hours
of observations. It consists of deep
SINFONI, seeing-limited, integral feld
spectroscopy of about 30star-forming
galaxies (Lyman-break selected), most of
which are at 3.0<z<3.7 and a few them
at 4.2<z<5.2. In the following we will
focus on the sample at z~3.3. For these
galaxies the nebular lines [OII] 3727
and [Ne III] 3869 are redshifted into the
H-band, while H and [OIII] 5007 are
redshifted into the K-band. The fux ratio
of these lines allows us to measure the
gas metallicity, as discussed in Maiolino
et al. (2008).
LSD (Lyman-break galaxies Stellar popu-
lation and Dynamics) is a companion
programme that was awarded 70 hours
of observations with SINFONI, assisted
by the adaptive optics module, so as
to achieve a much higher angular resolu-
tion relative to the seeing-limited obser-
vations. The sample consists of eight
Lyman-break galaxies at z~3.3 selected
to have a nearby bright star, which is
re quired to guide the adaptive optics sys-
tem.

In both projects the line emission (espe-
cially the strongest one, [OIII] 5007 )
is generally spatially resolved by our data.
Roberto Maiolino
1
Filippo Mannucci
2
Giovanni Cresci
2
Alessio Gnerucci
3
Paulina Troncoso
1
Alessandro Marconi
3
Francesco Calura
4
Andrea Cimatti
5
Filomena Cocchia
1
Adriano Fontana
1
Gianluigi Granato
6
Andrea Grazian
1
Francesca Matteucci
7
Taro Nagao
8
Laura Pentericci
1
Antonio Pipino
9
Lucia Pozzetti
10
Guido Risaliti
2
Laura Silva
6
1
Roma, Monte Porzio Catone, Italy
2
INAFOsservatorio Astrofsico di Arce-
tri, Firenze, Italy
3
Dipartimento di Fisica e Astronomia,
Universit degli Studi di Firenze, Italy
4
Jeremiah Horrocks Institute for
Astrophysics and Supercomputing,
University of Central Lancashire,
Preston, United Kingdom
5
Dipartimento di Astronomia, Universit
di Bologna, Italy
6
Trieste, Italy
7
Dipartimento di Fisica, Universit di
Trieste, Italy
8
Graduate School of Science and
Engineering, Ehime University, Japan
9
University of California Los Angeles,
USA
10
Bologna, Italy
The metal content in galaxies provides
important information on the physical
processes responsible for galaxy for-
mation, but little was known for galax-
ies at z>3, when the Universe was less
than 15% of its current age. We report
on our metallicity survey of galaxies at
z>3 using SINFONI at the VLT. We fnd
that at z>3, low-mass galaxies obey
the same fundamental relation between
metallicity, mass and star formation rate
as at 0<z<2.5; however, at z>3 mas-
sive galaxies deviate from this relation,
being more metal-poor. In some of
these massive galaxies we can even
map the gas metallicity. We fnd that
galaxies at z~3.3 have regular rota-
tion, though highly turbulent, and
inverted abundance gradients relative
to local galaxies, with lower abun-
dances near the centre, close to the
most active regions of star formation.
Overall the results suggest that promi-
nent infow of pristine gas is respon-
sible for the strong chemical evolution
observed in galaxies at z>3.
The chemical enrichment that we ob -
serve in local galaxies has been pro-
duced by the nucleosynthesis of stars
formed over their cosmic lives. Such
enrichment has been modulated by the
infow of pristine gas (which both boosts
star formation and dilutes the gas metal-
licity), enriched gas outfows (driven by
the star formation activity itself or by
active galactic nuclei [AGNs]) and gas
exchange during galaxy merging events.
The shape of the initial mass function
of star formation also plays an important
role, since different stellar masses inject
into the interstellar medium different
amounts of chemical elements. Clearly,
the metal content of galaxies is an im-
portant tracer of their star formation his-
tory and of the main physical processes
involved in galaxy evolution. Indeed, the
metallicity of local and distant galaxies
is one of the most important tools to con-
strain galaxy evolutionary models.
Clear observational evidence of a
connection between the content of met-
als and star formation history is given
by the tight three-dimensional correlation
between metallicity, stellar mass and
star formation rate (dubbed Fundamental
Metallicity Relation, FMR, Mannucci et
al., 2010), as illustrated in Figure1. More
specifcally, the gas metallicity is ob -
served to increase as a function of the
stellar mass (at a given star formation rate
[SFR]) and to decrease with the SFR (at
a given stellar mass). Galaxies show a
very small metallicity scatter of 0.05 dex
around this surface and this suggests
that the bulk of galaxy formation occurs
through a smooth, long-standing equi-
librium between star formation, gas infow
and outfow. Distant galaxies, out to
z<2.5, appear to obey the same funda-
AMAZE and LSD: Metallicity and Dynamical Evolution
of Galaxies in the Early Universe
37
The projected spatial resolution is typically
about 5kpc (seeing ~0.6 arcseconds) for
the AMAZE data and about 1.5kpc for
the (nearly) diffraction-limited data in LSD
(point spread function [PSF] ~0.2 arcsec-
onds). Also in both samples extensive
multiband photometry (including Spitzer
data) allowed us to constrain the stellar
masses tightly. The star formation rate
is inferred by using both the H luminos-
ity and spectral energy distribution (SED)
broadband ftting, generally obtaining
consistent results.
A detailed description of these two pro-
grammes, as well as preliminary results,
is given in Maiolino et al. (2008) and in
Mannucci et al. (2009). Additional results
have been, or are being published in fve
additional papers (Gnerucci et al., 2010;
Cresci et al., 2010; Troncoso et al., 2010;
Troncoso et al., in prep.; Gnerucci et al.,
in prep.), while follow-up observations are
delivering additional results. It is beyond
the scope of this paper to give an exten-
sive overview of the various results. Here
we only show some of the main highlights
that have been obtained by these pro-
grammes so far.
Chemical upsizing at z > 3
The AMAZE and LSD samples span more
than two orders of magnitude in stellar
mass (M
*
~10
9
10
11
M
A
) and over an
order of magnitude in star formation rate
2.2 2.2
2.2
2.2
2.2
2.2
2.2
3.3
1.4
1.4
0.8
0.8
1
0
1
8
8.25
8.5
8.75
9
1
2

+

l
o
g
(
O
/
H
)
9.5
10
10.5
11
lo
g
(M
a
s
s)
8
8.25
8.5
8.75
9
1
2

+

l
o
g
(
O
/
H
)
1
0
1 9.5 10 10.5
11
log(Mass) log(SFR)
3.3
2.2
2.2
0.8
0.8
2.2
Figure 1. Two views of the fundamental metallicity
relation (FMR) between mass, star formation rate
and metallicity for local galaxies are shown. The
small points, with different colours, indicate different
star formation rates. Squares with error bars indi-
cate the average location of distant star-forming gal-
axies at different redshifts (the latter indicated by
the number associated with each point). The point at
z~3.3 deviating from the FMR was obtained with
the frst preliminary set of 17 galaxies from AMAZE
and LSD. From Mannucci et al. (2010).
Figure 2. Metallicity versus star formation rate in gal-
axies at 0<z<2.5 on the FMR (lines) and in galax-
ies at z~3.3 (squares) is shown for three different
stellar mass ranges. Within each panel different col-
ours give the metallicity values at exactly the same
average stellar mass for galaxies both at 0<z <2.5
and at z~3.3. Note that while the metallicity of low-
mass galaxies at z~3.3 is consistent with local gal-
axies, massive galaxies at z~3.3 are signifcantly
more metal-poor relative to their local counterparts.
From Troncoso et al. (2010).
0
8.0
9.5
9.0
1 2 3 3 1 2 3 1 2
|og(SFR} (
/yr}
|og

= 9.15 (
}
|og

= 9.85 (
}
~ 3.3
0 < < 25
|og

= 10.6 (
}
9.0 (
}
9.3 (
}
9.9 (
}
9.8 (
}
10.7 (
}
10.5 (
}
1
2

+

|
o
g
(
O
/
H
}
|og(SFR} (
/yr} |og(SFR} (
/yr}
38
z~3.3 in our sample, by measuring the
velocity shift of the brightest line, [OIII]
5007 . About 30% of the galaxies in our
sample show ordered rotational motions
(Gnerucci et al., 2010a). A few examples
of such rotating systems are shown in
Figure3, where the central panels show
the rotation curve, along with the [OIII]
fux map (leftmost panels).
In relation to the chemical upsizing result
discussed above, we fnd that there is no
correlation between the dynamical prop-
(SFR ~30300 M
A
/yr). Therefore, the
metallicity inferred from the SINFONI
spectra allows us to obtain information
on the massSFRmetallicity relation
(FMR) at z~3.3. Figure1 shows the loca-
tion of galaxies at z~3.3 obtained
from the average of the frst set of 17gal-
axies observed in AMAZE and LSD,
showing that galaxies at z~3.3 clearly
deviate from the FMR.
In Figure 2 (from Troncoso et al., 2010a)
we exploit the full AMAZE and LSD
fnal samples. The solid and dashed lines
show a cut of the FMR (i.e. the metal-
licitySFR relation) at three different val-
ues of stellar mass. The AMAZE+LSD
data at z~3.3 are shown with solid
squares that, for sake of clarity, give the
average of the data in bins of mass and
SFR. Low-mass galaxies (M
*
~10
9.2
M
A
,
leftmost panel) have a metallicity in
line with the expectation of the relation
observed at 0<z<2.5, indicating
that these low-mass galaxies at z~3.3
are very much like local galaxies and
suggesting that they are regulated by the
same evolutionary processes. However,
massive galaxies at z~3.3, especially
at M
*
~10
10.7
M
A
(rightmost panel), are
signifcantly more metal-poor than galax-
ies at 0<z<2.5 with the same SFR.
Taken at face value, this result seems to
imply that massive galaxies at z~3.3
are in an earlier evolutionary stage rela-
tive to their low-mass counterparts, in the
sense that they have still to reach the
metallicitymassSFR relation character-
ising galaxies at low-z. This is in con-
trast with the expectations of downsizing
scenarios, where massive galaxies
should evolve faster and at earlier epochs
relative to low-mass systems.
There are a few possible scenarios
that could explain the deviation of mas-
sive galaxies at z~3.3 from the FMR
observed at 0<z<2.5. An excess of
pristine cold gas infow in massive galax-
ies, at such early epochs, may signif-
cantly dilute the gas metallicity. Alterna-
tively, galaxy mergers may drive gas
from the outer, low metallicity regions into
the central parts of massive galaxies,
hence diluting the metallicity of star-
forming regions. The latter scenario can
be investigated by studying the dynam-
ical properties of these systems, as in -
ferred by our SINFONI data.
Massive rotating discs at z > 3
The two-dimensional spectroscopic infor-
mation delivered by SINFONI allows us
to trace the kinematics of the galaxies at
0.0 0.2 0.4
Relati ve Intensity
0.6
0.7
0.5
0.5
1.0
1.5
0.0
0.0
R

(
)
[OIII] Flux
0.7
0.8 1.0 200 300 100 0
V (km/s)
100 200 300 7.5 8.0 8.5
12 + log (O/H)
9.0
[OIII] Velocity Metallicity
0.7
0.5
0.5
1.0
1.5
0.0
0.0
0.7 0.7
0.5
0.5
1.0
1.5
0.0
0.0
0.7
SSA22aM38
0.0 0.2 0.4
Relati ve Intensity
0.6
0.7
0.5
0.5
1.0
1.0
0.0
0.0
R

(
)
[OIII] Flux
0.7 1.4
0.8 1.0 200 100 0
V (km/s)
100 200 7.5 8.0 8.5
12 + log (O/H)
9.0
0.7
0.5
0.5
1.0
1.0
0.0
0.0
[OIII] Velocity
0.7 1.4 0.7
0.5
0.5
1.0
1.0
0.0
0.0
Metallicity
0.7 1.4
SSA22aM38
0.0 0.2 0.4
Relati ve Intensity
0.6
0.7
0.5
0.5
1.0
1.0
0.0
0.0
R

(
)
[OIII] Flux
0.7 1.4
0.8 1.0 200 100 0
V (km/s)
100 200 7.5 8.0 8.5
12 + log (O/H)
9.0
[OIII] Velocity Metallicity
0.7
0.5
0.5
1.0
1.0
0.0
0.0
0.7 1.4 0.7
0.5
0.5
1.0
1.0
0.0
0.0
0.7 1.4
SSA22aM38 Figure 3. [OIII] fux, velocity feld and metallicity maps
for three massive galaxies at z~3.3 characterised
by regular rotation patterns. The metallicity has a
minimum close to the central peak of star formation
(as traced by the maximum of H emission). From
Cresci et al. (2010).
Maiolino R. et al., AMAZE and LSD
39
erties of massive galaxies at z~3.3 and
their deviation from the fundamental
metallicitymassSFR relation observed
at 0<z<2.5. More specifcally, among
the z~3.3 massive galaxies, which
are metal-poor relative to the metallicity
massSFR relation at 0<z<2.5, we
fnd systems with both irregular kinemat-
ics (likely tracing merging systems) and
a regular rotation curve, in equal num-
bers. Therefore, merging cannot be the
only process responsible for lowering
the gas metallicity in these early galaxies.
From the rotation curve we can also infer
dynamical masses, which are in the
range 2 10
9
2 10
11
M
A
. Clearly, some
massive rotating discs are already in
place at this early epoch in the Universe.
However, in contrast to local disc galax-
ies, at z~3.3 rotating discs are much
more turbulent. Indeed, as illustrated in
Figure4, the velocity dispersion () is
generally comparable with the rotational
velocity (V
max
). More specifcally, the
average ratio between rotational velocity
and velocity dispersion is <V
max
/>
z = 3.3

= 2.2, to be compared with the value
<V
max
/>
z
= 0
= 10 typical of local discs.
Galaxies at z~3.3 appear to be even
more turbulent that those investigated at
z~2 by the parallel SINFONI programme
SINS (Frster Schreiber et al., 2009), which
are characterised by <V
max
/>
z = 2
= 4.5.
The highly turbulent nature of z~3.3
discs is likely due to very high gas
fractions, which make the discs dynami-
cally unstable. In samples at 1<z<2.5
high gas fractions have been con-
frmed directly through CO observations
(Daddi et al., 2010; Tacconi et al., 2010).
At z~3.3 we have obtained indirect
evidence for high gas fractions (even
approaching f
gas
=M
gas
/M
tot
~0.9) based
on the high surface density of star for-
mation (hence exploiting the Schmidt
Kennicutt relation) and on the comparison
between dynamical and stellar masses
(Mannucci et al., 2009; Gnerucci et al.,
2010; Troncoso et al. in preparation).
Such high gas fractions are likely associ-
ated with the prominent cold infows of
gas predicted to occur at such early
epochs by some theoretical models (e.g.,
Dekel et al., 2009).
Metallicity gradients and cold fows at
z>3
In some galaxies we not only resolve the
[OIII] 5007 emission (used to trace
the kinematics), but also the fainter lines
of [OII] 3727 and H, therefore en-
abling us to map the metallicity. The
rightmost panels in Figure3 (from Cresci
et al., 2010) show the metallicity map for
three massive galaxies characterised
by regular rotation velocity felds. Surpris-
ingly, in contrast to local galaxies, the
minimum metallicity is located close to
the central regions. However, the most
interesting result is that the minimum
metallicity is associated with the peak of
H fux, which traces the most active
star-forming regions. This result supports
the scenario where such massive sys-
tems at z~3.3 drive major infows of
pristine gas towards their central regions.
Such pristine gas both boosts star for-
mation and locally dilutes the pre-existing
medium, therefore producing the
observed spatial anticorrelation between
star formation and gas metallicity.
The same (strong) cold fow scenario can
explain, more generally for massive
galaxies at z~3.3, their reduced metal-
licity relative to the fundamental metallic-
itymassSFR relation observed at
0<z<2.5, as well as the highly turbulent
nature of these systems.
References
Cresci, G. et al. 2010, Nature, 467, 811
Daddi, E. et al. 2010, ApJ, 713, 686
Dekel, A. et al. 2010, Nature, 457, 451
Frster Schreiber, N. M. et al. 2009, ApJ, 706, 1364
Gnerucci, A. et al. 2010, A&A, submitted,
arXiv 1007.4180
Maiolino, R. et al. 2008, A&A, 488, 463
Mannucci, F. et al. 2010, MNRAS, 408, 2115
Mannucci, F. et al. 2009, MNRAS, 398, 1915
Tacconi, L. J. et al. 2010, Nature, 463, 781
Troncoso, P. et al. 2010, A&A, submitted
Figure 4. Velocity dispersion () versus rotational
velocity (V
max
) is shown for disc galaxies at z~3.3
with a regular rotation pattern. The solid line shows
the relation V
max
/=10 typical of local galaxies. Note
that discs at z~3.3 are much more turbulent than
local galaxies, many of them having V
max
/~1. From
Gnerucci et al. (2010).
0
0
50
100
150
200
ODFS-9340 ODFS-2528 SSA22A-M38
O0302-O131
ODFS-16767
AMAZE objects
|SD objects
SSA22A-D17
ODFA-O9
ODFS-14411
ODFS-9313
SSA22A-O16
3O324-O3
400
(km/s}
|n
t

(
k
m
/
s
}
= 10
|nt
= 7
|nt
= 4
|nt
= 2
|nt
= 0.25
|nt
= 0.5
|nt
=
|nt
|
o
c
a
| g
a
|a
x
|e
s
S
l
N
S

g
a
|
a
x
|
e
s

(

~

2
}
600
A
s
t
r
o
n
o
m
i
c
a
l

N
e
w
s
His Majesty King Harald of Norway (far right)
presents the 2010 Kavli Prize in Astrophysics to,
from left to right, Jerry Nelson, Raymond Wilson
and Roger Angel.
The inauguration of the EVALSO (Enabling Virtual
Access to Latin-American Southern Observatories)
high speed network project held at ESO Offces
in Santiago Chile on 4 November 2010. See the
article by Filippi (p. 2).
C
r
e
d
i
t
:

S
c
a
n
p
i
x
41
Astronomical News
Jeremy Walsh
1
1
ESO
Ray Wilson, who retired from ESO in
1993, was awarded two prestigious
prizes in September 2010 for his out-
standing work on telescope optics: the
Kavli Prize in the feld of astrophysics
and the Tycho Brahe Prize 2010 of the
European Astronomical Society.
Announced in June 2010, and awarded
in Stockholm in September, the million-
dollar Kavli Prizes were awarded to
eight scientists whose discoveries have
dramatically expanded human under-
standing in the felds of astrophysics,
nano science and neuroscience (see the
Kavli press release
1
). The Kavli prize is
awarded by the Norwegian Academy of
Science and Letters, the Kavli Foundation
and the Norwegian Ministry of Educa-
tion and Research. The Kavli Foundation
is funded by Fred Kavli, the Norwegian
entrepreneur and philanthropist who later
founded the Kavlico Corporation in
the US today one of the worlds largest
suppliers of sensors for aeronautical,
automotive and industrial applications.
There were three recipients in astro-
physics, all acclaimed for their work on
the development of giant optical tele-
scopes
2
Roger Angel of the University
of Arizona, Tucson, attached to Steward
Observatory; Jerry Nelson of the Uni-
versity of California, Santa Cruz and long
associated with the Keck Observatory;
and Ray Wilson. The photograph on
the facing page (upper) shows the three
prize winners at the prize award cere-
mony in Oslo.
The Tycho Brahe Prize of the European
Astronomical Society (EAS) is awarded
annually in recognition of the develop-
ment or exploitation of European instru-
ments, or major discoveries based large-
ly on such instruments. The prize is
sponsored by the KlausTschira founda-
tion, based in Heidelberg, Germany. An-
nounced in April, Ray received the prize
at the JENAM meeting in Lisbon, Portu-
gal (see the article by Sandu and Chris-
tensen, p. 42) and, at the plenary session
on 10 September 2010, he delivered a
lecture entitled From the ESO NTT to the
VLT and the 42-metre ELT: the develop-
ment of active optics as the basis of
all modern telescope optics. Figure 1
shows Ray receiving the Tycho Brahe
prize from the retiring president of the
EAS, Joachim Krautter.
Ray Wilson, who was born in England
and educated at Birmingham University
and Imperial College London, arrived at
ESO in 1972 from Zeiss at Oberkochen,
Germany where he had been head of the
Optical Design Department for Astro-
nomical and Analytical Instruments. At
ESO, frst in Geneva and then in Garching,
he was the frst head of the ESO Optics
and Telescopes Group. The revolutionary
active optics of the 3.58-metre NTT, in-
spired by two years of work at the La Silla
Observatory, was the crowning achieve-
ment of Rays work at ESO. This success-
ful concept, where both the alignment of
the optical elements as well as the shape
of the fexible primary mirror are con-
trolled in a closed loop based on the
measurements of a wavefront sensor,
was then also used for the VLT 8.2-metre
telescopes. In addition, Ray contributed
to telescope designs with more than
two powered mirrors, which are now
being explored for the next generation of
extremely large telescopes, such as
the European Extremely Large Telescope
(E-ELT) project.
During his last years at ESO he began
work on his magnum opus, the two-vol-
ume work Refecting Telescope Optics,
published by Springer
5
. Volume I: Basic
Design Theory and Its Historical Develop-
ment, frst appeared in 1996, and Volume
II: Manufacture, Testing, Alignment, Mod-
ern Techniques followed in 1999. Both
are currently in their second edition and
Ray is working on updates for the third
editions. Ray has also been honoured by
a number of other prizes, including the
Karl Schwarzschild Medal of the German
Astronomical Society, an appointment as
Chevalier of the French Lgion dHonneur
and the Prix Lallemand of the French
Academy of Sciences.
Links
1
The Kavli Prize: http://www.kavlifoundation.org/
kavli-prize
2
Announcement of Kavli Prize for astrophysics:
http://www.kavlifoundation.org/2010-astrophysics-
citation
3
ESO announcement of Kavli Prize:
http://www.eso.org/public/news/eso1022/
4
Announcement of Tycho Brahe Prize:
http://www.eso.org/public/announcements/
ann1018/
5
Wilson, R. N. 2004, Refecting Telescope Optics,
Volumes I and II, 2nd edition, (Heidelberg: Springer)
Raymond Wilson Honoured with Two Prestigious Prizes
Figure 1. Ray Wilson receiving the Tycho Brahe prize
from the president of the EAS at the JENAM meeting
in Lisbon, Portugal.
42
Astronomical News
(e.g., introduction to the data reduction
system, etc.). Regrettably ESO will
not be able to cover any expenses (travel,
ac commodation, etc.) for these data
reduction missions. The visitor will use
her/his own laptop to run the data reduc-
tion software remotely via the standard
GUIs. The raw data will then be trans-
ferred to the reduction machine, either
from a laptop (i.e. via ftp) or from the
archive ftp site after an archive request
has been processed. Saving the reduced
data is the responsibility of the user.
Visitors are expected to spend at most
fve days on each data reduction mission,
and the service will be available during
normal offce hours from Monday to
Friday at ESO Headquarters in Garching.
Users wishing to employ this HARPS re-
duction service in Garching should send
an email to re-harps@eso.org.
Links
1
Details of the service at: www.eso.org/sci/facilities/
lasilla/instruments/harps/tools/reprocess.html
Tamai, Head of ESOs Technology Divi-
sion, showcased the technology of the
VLT/VLTI, while Roberto Gilmozzi, Head
of ESOs Telescope Division, presented
the principal technological features of the
European Extremely Large Telescope
(E-ELT); a project that incorporates many
innovative developments. A presentation
on control software and data reduction
and analysis was delivered by Michle
Pron, Director of Engineering and Soft-
ware Development.
On Tuesday, 7 September, there was a
dedicated ESO plenary session. Bruno
Leibundgut gave a comprehensive talk
about recent developments at the La Silla
Paranal Observatory, including plans for
Gaspare Lo Curto
1
Thierry Benifah
1
Andrew Burrows
1
Eric Emsellem
1
Kevin Maguire
1
Luca Pasquini
1
John Pritchard
1
Martino Romaniello
1
1
ESO
From the start of the year 2011 the HARPS
data reduction software will be also avail-
able at ESO Headquarters in Garching.
This new initiative will enable users to ap-
ply for access to the system locally in
Garching.
The experience of the past few years of
operations of HARPS has shown that
sometimes the data reduced online at the
La Silla Observatory might require further
reprocessing and analysis. This is usually
because the wrong set of initial parame-
ters have been specifed in the observing
template (such as stellar spectral type or
Oana Sandu
1
Lars Lindberg Christensen
1
1
ESO
The Joint European and National Astron-
omy Meeting (JENAM) that took place
in Lisbon, Portugal, during the week of
610 September 2010, was the 18th
Annual Meeting of the European Astro-
nomical Society (EAS) and the 20th
Annual Portuguese Meeting of Astronomy
and Astrophysics. JENAM brings Euro-
pean astronomers together to discuss
frontline topics in astronomy, space sci-
ence and instrumentation technology.
initial guess for the radial velocity). In
these cases the spectral extraction is not
affected, but the precision and the accu-
racy of the radial velocity measurement
are generally not optimal. Although these
cases are not frequent, they do happen
from time to time and require re-compu-
tation of the radial velocities. We aim to
address such needs by allowing indi-
vidual users to visit ESO Headquarters in
Garching and give them access to the
same data reduction software that is
available at the La Silla Observatory site,
both in its on- and off-line favours.
Users wishing to take advantage of this
service are encouraged to check the
details at the web page
1
. The user should
then contact ESO giving a brief scien-
tifc and technical rationale as to why
reprocessing is required, together with
the amount of data that needs to be
reduced and the intended dates of travel
to Garching. ESO, after checking availa-
bility, will make available desk space and
will grant access from the users laptop
to the data reduction computer. Limited
on-site user support will also be provided
ESO was extensively involved in the
meeting, highlighting its role as a driving
force in ground-based astronomy at the
European level, as well as globally. Sev-
eral key ESO people participated in
the meeting and there were also an ESO
plenary and special session, an ESO
exhibition with free educational and infor-
mational material and a book launch.
During the frst day of the meeting, ESO
participated in the special session on
Astronomy Challenges for Engineers and
Computer Scientists with talks by Bruno
Leibundgut, Director for Science, on
science projects at ESO and Andreas
Kaufer, Director of La Silla Paranal Obser-
vatory, on ESOs infrastructures. Roberto
Availability of Reduction Software for HARPS Data
at ESO Headquarters in Garching
ESO Participation at the Joint European and
National Astronomy Meeting in Lisbon, Portugal
43
the E-ELT. He was followed by the lead-
er of the ESO Survey Team, Magda
Arnaboldi, who presented the status of
the Public Surveys that have recently
begun with VISTA.
A special session on ALMA Early Science
took place on the same day. The session
included presentations on the current
status of the construction project, the
ALMA development plan and its opportu-
nities, and the European ALMA Regional
Centre plans for user support in prepa-
ration for Early Science. The session also
included a demonstration of the ALMA
software that will be used to apply for ob-
serving time on ALMA, to prepare obser-
vations, interact with the ALMA archive
and reduce the data.
ESO also participated in the session ded-
icated to Education and Outreach after
IYA2009 in Europe. Pedro Russo, Global
Coordinator for the International Year of
Astronomy 2009, talked about his experi-
ence in spearheading this project.

During the fve days of the conference,
ESO exhibited its astronomical dis-
coveries, and the telescopes that made
these discoveries possible. The E-ELT
project attracted the interest of many
passers-by who admired the telescope
model and went on to learn more about
the instrument that will be the worlds
biggest eye on the sky.
In order to encourage interaction be-
tween participants and foster debates on
topics related to astronomy in a more
relaxed environment, ESO ran the ESO
Hour at its stand on three days of the
meeting. The event proved to be suc-
cessful, gathering numerous participants
from the meeting who engaged in con-
versation with people from the organisa-
tion.
One of the highlights at the ESO stand
was the launch of the book Postcards
from the Edge of the Universe. This book
is based on the science carried out by
a hand-picked selection of the best blog-
gers from the Cosmic Diary
1
, one of the
12 Cornerstone projects of the Interna-
tional Year of Astronomy 2009. Twenty-
four astronomers from all corners of
the globe explain their science in articles
edited by Lee Pullen, Mariana Barrosa
and Lars Lindberg Christensen.
Announced several days in advance
and kept as a surprise, the book launch
took place on Tuesday, when several
of the authors and editors talked about
the book (see Figure 1). A special talk
was given by a guest of honour the
former ESO Director General Catherine
Cesarsky. A book signing session took
place and free postcards signed Greet-
ings from the Edge of the Universe
were distributed to people, who were
encouraged to post them to their family
and friends. More infor mation on obtain-
ing PDF copies of the book, or infor-
mation on how to order a hard copy or
send electronic postcards, are available
2
.
Among other highlights of the meeting
included the inauguration of the EAS
Lodewijk Woltjer Lecture award talks with
Woltjer himself giving the frst talk. Under
his leadership as Director General, ESO
signed the VLT contract and established
itself as one of the worlds leading astro-
nomical institutes. Lodewijk Woltjer also
made signifcant contributions to theore-
tical astrophysics from his fundamental
work on the Crab Nebula and his studies
on the energy source of radio galaxies
and quasars. Another highlight of the
JENAM meeting was the award of the
EAS Tycho Brahe Prize for 2010 to tele-
scope designer Ray Wilson (see the arti-
cle on p. 41).
Links
1
Cosmic Diary: http://www.cosmicdiary.org
2
Postcards from the Edge of the Universe:
http://www.postcardsfromuniverse.org
Figure 1. The ESO stand at JENAM in Lisbon, Portu-
gal shown during the launch of the the book Post-
cards from the Edge of the Universe. Catherine
Cesarsky, former ESO Director General and past
President of the IAU, is shown discussing the book.
44
Astronomical News
Michael West
1
1
ESO
Astronomers travelling to Chile to observe
with ESO telescopes are encouraged to
include a visit to ESOs offce in Santiago.
With the recent completion of the new
ALMA building (see Figure 1), there is now
a vibrant scientifc community of more
than 100 as tronomers, fellows and stu-
dents at the ESO premises in Santiago.
ESOs Offce for Science in Chile is happy
to provide accommodation and per diem
for visiting astronomers who wish to give
a science talk and interact with ESO and
ALMA staff. If interested, please contact
Michael West (mwest@eso.org). We look
forward to welcoming you to Santiago!
Visiting ESOs Offce in Santiago
Figure 1. The hand-over ceremony for the new
ALMA offces at the ESO premises in Santiago. Tim
de Zeeuw (ESO Director General, left) and Thijs
de Graauw (ALMA Director, right) exchange the key.
Henri Boffn
1
Hannelore Hmmerle
2
Barbara Wankerl
3
Silke Zollinger
4
1
ESO
2
Max-Planck-Institut fr extraterrestrische
Physik, Garching, Germany
3
Excellence Cluster Origin and Structure
of the Universe, Garching, Germany
4
Max-Planck-Institut fr Physik, Munich,
Germany
On a Monday evening such as the 5th of
July this year, the trendy bar Caf Jasmin
in the Schwabing district of Munich is
full inside, despite the hot air and blazing
Sun that would normally invite people
to enjoy a beer outside in a tree-shaded
Caf & Kosmos Events in Munich
Figure 1. An interested crowd in lively discussion
with the physicist Stefan Stonjek from MPP at the
frst Caf & Kosmos event.
C
r
e
d
i
t
:

A
.

G
r
i
e
s
c
h
/
M
P
P
45
garden. Although we are in the midst
of the international football season, these
people did not gather to watch another
game of the World Cup; oddly enough,
these are members of the public inter-
ested in science, and in our Universe in
particular. They are here to listen and
discuss with ESO astronomer Markus
Kissler-Patig, the fundamental question:
Are we alone in the Universe?
Subsequent science cafs have con-
frmed the success of the frst event
that took place on 31 May 2010, when
Stefan Stonjek, a physicist from the
Max-Planck-Institut fr Physik (MPP),
discussed The Big Bang in the Tunnel,
covering the latest details about the
CERN Large Hadron Collider (see Fig-
ure 1). The complex topic of string theory
was aired by Ilka Baumgartl and Marco
Baumgartl from the Excellence Cluster in
September and the subject of the black
hole at the heart of the Milky Way was
discussed by Stefan Gillessen from the
Max-Planck Institute for Extraterrestrial
Physics in November. The series is enti-
tled Caf & Kosmos and is a joint initiative
between ESO, the Excellence Cluster
Origin and Structure of the Universe, and
the Max-Planck Institutes for Physics,
Astrophysics and Extraterrestrial Physics.
The idea is to bring science directly to
the general public in the Munich area in
the relaxed atmosphere of a bar.
For many, research is far away and
happens behind the closed doors of
laboratories. Admittedly, there are many
science magazines in print or on televi-
sion and some newspapers have a page
about science, but the contact between
science and the public is often very in -
direct. Many scientifc institutions organ-
ise open days, which are often very
successful, but these take place only
once a year at most. Public conferences
on scientifc themes are also often popu-
lar, but they generally follow the same
academic scheme: the scientist speaks
and the public listens. Direct exchanges
between scientists and the public seldom
take place.
The Caf & Kosmos initiative aims to
bring researchers and non-scientists
together, and to do so in places where
people typically meet, share their
thoughts, discuss business and debate
about big and small things. Thus a pub
in the centre of the city of Munich was
chosen for the meetings; a place where
communication traditionally takes place.
With Caf & Kosmos, we want to give
people the chance to speak directly with
scientists about current fascinating sci-
entifc themes.
The proposed themes for Caf &
Kosmos come from astrophysics, cos-
mology and particle physics, and are
on topics of great interest for non-scien-
tists, such as What are black holes?,
What do we know about dark matter?,
Why did CERN build the LHC?, What
do we know about planets outside our
Solar System?, and so on. The discus-
sions are held in German. These and
other topics are discussed in the relaxed
atmosphere of a pub every frst Mon-
day of the month. The duration of the
discussions is initially about one and a
half hours, although our frst two experi-
ences have shown us that many people
tend to stay for much longer.
For more information on the Caf &
Kosmos series, including the
list of future speakers, please go to
http://www.cafe-kosmos.org.
It is a pleasure to thank Aleks Vulic,
owner of Caf Jasmin, for his permission
to use his premises for these interactive
sessions, as well as the speakers.
my passion for history and astronomy
occurred in a remote Spanish village
when I was approximately ten years old.
My uncle had a TV (our family did not)
and I watched an episode of the well-
known science fction series Cosmos
1999. It came as a revelation that
man could travel into space and reach
remote parts of the Universe. I imme-
diately became interested in rockets and
space shuttles. Finding and studying
new civilisations seemed to be a realistic
project and it took me a few years
to accept that current technology was
Jean-Philippe Berger
I have been at ESO as a VLTI staff as -
tronomer, on leave from the Laboratoire
dAstrophysique de Grenoble (France),
for about 10 months. I came to Santiago
with my wife Stephanie and our three
children Clara, Lucie and Axel.
I remember my very early fascination with
ruins that later turned into a deep inter-
est in human civilisations and especially
their dawn and dusk. This has never
left me. I believe the connection between
New Staff at ESO
Astronomical News
barely capable of fying to the planets in
the Solar System. Then at the age of 13 I
received a book from my uncle: Le Ciel
by Jean-Claude Pecker. This was the
true revelation that one could travel in
space thanks to telescopes and an imag-
inative brain.
Even though I continued to follow his-
tory lectures throughout my academic
career, my educational path slowly
but surely shifted towards science and
technology and the fnal call to become
an astronomer came during my short
46
Astronomical News
Universit Joseph Fourier and in public
outreach and realised that it was an es -
sential part of our mission as astron-
omes de la Rpublique. Finally, in 2008,
at a conference in Marseille, some of
my colleagues and myself were impressed
by the imaging results coming out from
the MIRC instrument at CHARA and
became convinced that the VLTI and
integrated-optics technology were ready
for four-telescope operation. This dis-
cussion gave birth to the PIONIER VLTI
visitor instrument project that I initiated
and led in collaboration with some in
December 2008. Since then PIONIER has
been built and it received its frst stellar
photons in October 2010.
Coming to ESO and VLTI was thus a
natural move. First I could follow PIONIER
and secondly I was keen to join the
impressive Paranal astronomical machine
and the excellent VLTI team.
stay at ENS Lyon for my masters in astro-
physics. After a two-year stay in Chad,
Africa, as a physics teacher, I joined the
Laboratoire dAstrophysique de Grenoble
(LAOG) in 1995 for a PhD on the polar-
ising properties of dusty environments
in young pre-main sequence binaries.
Towards the middle of my PhD I realised
that I wanted to become involved in an
instrumentation project. One day, on the
stairway, I met a colleague who was
also an optical engineer, Pierre Kern, who
was very excited by his new idea to use
the miniature optical circuits developed
by the telecommunications industry to
produce interference between the light
of two telescopes. I was immediately
struck by the potential of this technology
for aperture synthesis in the optical
and asked him if there was a way for me
to contribute to the project. Unfortunately
he was applying for a grant to recruit a
PhD student and had a good candidate;
but fortunately (for me) the grant never
arrived and I spent the remaining part of
my PhD developing this new and prom-
ising technology. Since then the term
Astrophotonics has been coined to
describe the marriage between photonics
and astrophysics.
So, eighteen months from the end of my
PhD grant I started everything from
scratch, and I found myself alone in an
empty lab with the task of exploring
how to apply photonic technology to
astronomical interferometry. Fortunately I
married Stephanie almost at the same
time as I changed my research path and
her presence by my side for all these
years has been essential in holding me to
this new path. My thesis culminated in
resolving the accretion disc of a young
star, FU Orionis, for the frst time with
astronomical unit resolution, working with
my unoffcial but remarkable advisor
and friend Fabien Malbet, who taught me
how to use the Palomar Testbed Inter-
ferometer, which was then the only instru-
ment with suffcient sensitivity to study
protoplanetary discs. At the end of my
PhD I realised that my research interests
combined challenging astrophysics with
challenging instrumentation.
I continued developing integrated optics
technology after my thesis, thanks to
the support of the French Space Agency
(CNES). I dived into the photonics world
for two years at LEMO and in 2000 I
obtained a NASA/JPL Michelson Fellow-
ship to work with Wesley Traub and his
team at the HarvardSmithsonian Center
for Astrophysics. My project was to install
an instrument, IONIC3, capable of com-
bining three beams of the IOTA interfer-
ometer in Arizona and to use it to gener-
ate the frst images of the circumstellar
environment in a pre-main sequence star.
I remember those American years with
a lot of affection, and the seemingly
countless days and nights spent at IOTA
(Mount Hopkins, Arizona) and CHARA
(Mount Wilson) still live with me. We
expended a huge amount of energy to try
to produce the frst maps of young stellar
objects in the near-infrared. This was
partially successful, but clearly limited by
the lack of long baselines at IOTA.
In 2002 I took up a permanent position
at the Observatoire de Grenoble, where
I continued to develop new projects
related to astronomical aperture synthe-
sis imaging. I also started to get more
and more involved in teaching at the
Jean-Philippe Berger
47
Margaret Moerchen
I grew up in central Texas, where city
lights frequently obscured the nightly
show of constellations. (While we proudly
refer to Texas as The Lone Star State,
it isnt for this reason!) When it was time
for a meteor shower, I could make a
short drive into the hill country where a
much richer sky was revealed and
perhaps its because the stars werent
always in plain sight that these glimpses
were so inspiring.
These night-time excursions continued
while I did my undergraduate course-
work at the University of Texas at Austin.
I then found an even darker sky in the
Davis Mountains, home of McDonald
Observatory, where I had my frst experi-
ence using a big (0.75-metre) telescope.
After graduating, I took my frst step
into mid-infrared astronomy by working
on the design of EXES (Echelon Cross
Echelle Spectrograph), a spectrograph
that will soon fy on SOFIA, the Strato-
spheric Observatory for Infrared Astron-
omy.
Wanting to continue in the hot feld of the
thermal infrared, I went to the University
of Florida to participate in the building of
mid-infrared instruments such as T-ReCS
(Thermal Region Camera and Spectro-
graph) for Gemini South. I spent fve
monsoon seasons in Gainesville, where
my thesis research employed T-ReCS
and other mid-infrared cameras to study
the architecture of young planetary sys-
tems by determining the location of warm
dust within them. At the same time, I had
the opportunity to become involved in
the development of CanariCam (presently
in the commissioning phase), the mid-IR
facility instrument for the Gran Telesco-
pio Canarias, the 10.4-metre segmented-
mirror telescope on La Palma.
After attending the Observing Planetary
Systems workshop at ESO Chile in 2007,
I thought this would be a fantastic place
to combine the pursuits of both scientifc
research and instrumentation develop-
ment. I became even more sold on ESO
after I had the chance to visit Paranal
to assist in a post-intervention characteri-
sation of VISIR. Now, almost three years
later, Im back at the VISIR console as
instrument fellow and am part of the team
working on its exciting upgrade project.
My functional duties include support of
observations not only with VISIR, but with
all instruments at Unit Telescope 3 (UT3)
and UT4, and at UT4 Ive been able to
learn more about adaptive optics and the
near-infrared regime. Performing a wide
variety of science programmes at the two
telescopes has provided a unique per-
spective on the power of 8-metre tele-
scopes, and welcoming visiting astrono-
mers and hearing enthusiastic detailed
descriptions of their projects is one of the
most rewarding aspects of the support
work. In fact, even driving astronomers
up or down the mountain in the middle of
the night offers one of the benefts we
sometimes forget to look at the night
sky! Ive seen the most spectacular
crumbling-freball meteors of my life while
on the summit road, and theyre an im -
pressive reminder of why I travelled thou-
sands of miles to stay up at night in the
middle of the desert.
Davor Krajnovi
On a warm summer night, in the com-
pany of a dentist, a physician and a man-
ager, in the cockpit of a sailing boat that
rocks slowly, while a breeze brings the
smell of pine trees and the buzzing sound
of a few persistent cicadas, and the stars
shine on us with the intensity of a crystal-
clear night, the dentist concludes: What
a good job you have: looking at the stars!
I often ask myself, how many decisions
I made to get here. Until I was more than
half way through my university physics
degree I was not thinking of being a pro-
fessional astronomer. I had not even
looked through a telescope until about
that time. One of the frst convincing
moments was when I joined a group of
fellow students on a visit to the observa-
tory in the sleepy town of Vinjan in Croa-
tia, where amateur astronomers were
becoming professionals in their achieve-
ments of spotting Near Earth Objects.
There I had a frst glimpse of the life of an
astronomer: it was not just star-gazing.
There was a lot of careful and patient tak-
ing of pictures, comparing them, and
working with different image-processing
software, and so on, throughout the
night, until the rain came. It was fun!
The starting point in my career was the
acceptance to do a PhD at the Leiden
Observatory. Very early on, I went to
observe with the SAURON integral-feld
unit mounted on William Herschel Tele-
scope on La Palma. The ffth observing
run of the SAURON Project was 14 nights
long and a number of team members
came to share the time, but I stayed for
the full run. Two weeks on the mountain,
not having to close the dome once,
catching as many photons as the detec-
tors allowed, from dusk until dawn, and,
just before going to sleep, looking at the
shadow of the mountain on the clouds
or the Atlantic Ocean, this was something
extraordinary. I was hooked.
Margaret Moerchen
Fellows at ESO
Astronomical News
48
Astronomical News
I spent four wonderful years in Leiden.
My thesis consisted of the analysis of
the nuclear properties and dynamical
modelling of nearby early-type galaxies,
as well as developing a method to ana-
lyse the two-dimensional kinematic
maps coming from instruments such as
SAURON. After the PhD, I went to Oxford
where I stayed for fve years. For the
last two I was an Extraordinary Junior
Research Fellow of Queens College (the
Extra actually means they were not
paying my salary, but they did take care
of me as a college does). There I started
working in the near-infrared and with
laser-guided adaptive optics observa-
tions, which unfortunately also meant that
trips to the telescopes became rarer.
A large part of my science, however, con-
centrated on a new and exciting project,
the ATLAS3D survey, which I am co-
leading. This is a multi-wavelength survey
of a complete sample of nearby early-type
galaxies, and it includes a large team of
observers and theoreticians. The frst
results are coming out right now and it
is amazing to see how the initial ideas
have turned into science.
It is now about ten years since I left my
hometown and I am entering my sec-
ond year as an ESO fellow. Astronomy
really is not just star-watching, and ESO
is a prime example of the complexity
needed for successful astronomical oper-
ations. I feel rather privileged to be able
to participate in it. On a warm summer
night, however, when a friend says it is a
good job, this star watching, one has to
make a decision: to agree, or say: Well,
actually I study black holes ...
Tim de Zeeuw
1
1
ESO
Christine (Chris) Nieuwenkamp was born
in Belgium and studied at the Higher
Institute for Translators in Antwerp. She
started her professional career in the
purchasing department of a Brazilian min-
ing company stationed in Belgium. She
joined ESO in April 1990 as Administrative
Assistant for Purchasing in the Contracts
and Procurement Service at ESO Head-
quarters in Garching.
In Memoriam Christine Nieuwenkamp
Through her strong motivation and dedi-
cation, Chris was able to cope with
a constantly high workload, especially
related to calls for tender, contracts
and contract administration, and inter-
acted with a large number of staff in
both Garching and Chile. In addition
Chris successfully trained and integrated
new staff in the Contracts and Procure-
ment Department. For several years
Chris was an active and respected mem-
ber of the ESO Staff Association in
Garching. Her excellent performance as
a Procurement Offcer was well-recog-
nised both inside and outside ESO and
she was a pleasure to work with. She
died on 18 October 2010. Chris will be
remembered by her friends and col-
leagues at ESO and missed by her hus-
band and two children.
Davor Krajnovi
49
ALMA, the Atacama Large Millimeter/
submillimeter Array, is expected to be the
leading observatory at millimetre and
submillimetre wavelengths over the com-
ing decades. It is the result of a global
collaboration involving Europe, North
America, East Asia and the host country
Chile. When completed, it will com-
prise at least 66 high precision antennas
equipped with receiver and digital elec-
tronics systems to observe in the fre-
quency range from 30 GHz to 1 THz and
achieve angular resolutions as high as
5 milliarcseconds. Dynamic scheduling
and innovative calibration strategies
will ensure the most effcient use of the
unique atmospheric qualities encoun-
tered at the 5000-metre high site on the
Chajnantor plateau in the northern Andes.
While Full Science Operations are esti-
mated to begin in 2013, the increasing
capabilities of the growing array will
become available to the astronomical
community following the start of Early
Science Operations in the second half of
2011. During the frst phase of Early
Science, an array of 16 antennas will be
offered for interferometry with four fre-
quency bands and a limited range of
baselines. Early Science observations are
currently estimated to be scheduled for
at most one third of the available time,
the remainder being reserved for continu-
ing commissioning and science verifca-
tion activities.
Scientifc users will interact with the
ALMA facility through their local ALMA
Regional Centre (ARC), which will provide
user support on all aspects related to
observing with ALMA and assist observer
teams throughout the lifecycle of their
project. The European ALMA community
is supported by a network of regional
ARC nodes that are coordinated by
the central European ARC hosted at ESO
Headquarters in Garching, Germany.
With the ALMA Community Days, the
ESO ARC aims to prepare the European
astronomical community for ALMA
Early Science operations. The frst day
will be dedicated to a series of scientifc
and technical presentations related
to ALMA and Early Science capabilities,
while the second day will be taken up
by interactive tutorials on the preparation
of ALMA observing proposals using the
ALMA Observing Tool (OT). This should
help novice and advanced ALMA users
alike to create observing projects that
optimally exploit the unique capabilities of
ALMA during Early Science operations.
Further information can be found at
www.eso.org/sci/meetings/2011/
alma_es_2011.html or by emailing
alma_es@eso.org.
Announcement of the
Astronomical News
ALMA Community Days: Towards Early Science
67 April 2011, Garching, Germany
This workshop will provide an overview of
recent observational results on the stellar
systems in nearby galaxy clusters, i.e.
Fornax, Virgo, Coma et al., and a forum for
discussion and comparison of theoretical
models for the evolution of galaxies and
larger-scale structures with observational
properties of stellar systems in high den-
sity environments at redshift zero. The
aim is also to identify those questions that
can be tackled by the European Extremely
Large Telescope (E-ELT), as the excep-
tional high angular resolution and collect-
ing power of an extremely large telescope
are essential ingredients for the study
of resolved stellar populations at distances
larger than 10 megaparsecs. The nearby
clusters Fornax, Virgo, Coma et al. will
be the frst obvious targets of this exciting
new era in extragalactic astronomy.
Topics to be covered include:
1) the faint end of the galaxy luminosity
function in clusters: dwarf galaxies,
ultra-compact dwarfs and globular clus-
ters;
2) the bright end of the galaxy luminosity
function: stellar populations and dynam-
ics observation and theory;
3) surveys of nearby clusters and their fol-
low-up with the ESO Very Large Tele-
scope (VLT) and future facilities (E-ELT,
Atacama Large Millimeter/Submillimeter
Array, ALMA); and
4) the 3D structure of nearby galaxy clus-
ters and the morphological transforma-
tion of galaxies.
The number of participants is limited
to the capacity of the ESO auditorium
(about 90 participants).
Scientifc Organising Committee:
Magda Arnaboldi (chair), Nobuo Arimoto,
Michele Cappellari, Eric Emsellem, Bill
Harris, Ken Freeman, Ortwin Gerhard,
John Kormendy, Harald Kuntschner,
Claudia Maraston, Lucio Mayer, Simona
Mei, Bianca Poggianti, Sadanori Okamura,
Tom Richtler.
Local Organising Commitee:
Magda Arnaboldi, Lodo Coccato, Luca
Cortese, Michael Hilker, Marina Rejkuba,
Christina Stoffer.
Web page for registration and further
details will appear soon at http://www.eso.
org/sci/meetings.html.
Fornax, Virgo, Coma et al.:
Stellar Systems in Nearby High Density Environments
27 June1 July 2011, Garching, Germany
Announcement of the ESO Workshop
50
Personnel Movements
Arrivals (1 October31 December 2010)
Europe
Ascenso, Joana (P) Fellow
Beccari, Giacomo (I) Fellow
Boissier, Jrmie (F) Fellow
Bressert, Eli (USA) Student
Coccato, Lodovico (I) Fellow
Gabasch, Armin (I) Software Engineer
Heidecke, Thies (D) Student
Kains, No (B) Fellow
Lablanche, Pierre-Yves (F) Student
Longmore, Steven (GB) Fellow
Ltzgendorf, Nora (D) Student
Pettazzi, Lorenzo (I) Control Engineer
Chile
Alvarez, Fernando (RCH) Head of the Maintenance Department
Gonzalez, Jaime (RCH) Optical Technician
Jones, Matias (RCH) Student
Miccolis, Maurizio (I) System Engineer
Rodrigues, Myriam (P) Fellow
Vanderbeke, Joachim (B) Student
Vega, Florine (RCH) Procurement Clerk
Departures (1 October31 December 2010)
Europe
Benifah, Thierry (F) Head of Information Technology
Bois, Maxime (F) Student
Cortesi, Arianna (I) Student
Frank, Matthias (D) Student
Guidetti, Daria (I) Student
Hansen, Camilla Juul (DK) Student
Hewitson, Jennifer (D) Secretary/Assistant
Larsen, Jonas (DK) Software Engineer
Lombardi, Marco (I) Astronomer
Naets, Thomas (B) Internal Auditor
Neumayer, Nadine (D) Fellow
Penuela, Tania Marcela (CO) Student
Ruiz Velasco, Alma (MEX) Student
Russo, Pedro Miguel (P) Astronomer
Seemann, Ulf (D) Student
Taylor, Luke (GB) Laser Physicist
Unterguggenberger, Stefanie (A) Student
Williams, Michael (GB) Student
Chile
Asmus, Daniel (D) Student
Kaminski, Adrian (D) Student
Raff, Gianni (I) Head of the Computing IPT
Thomas, Alexis (RCH) Network Specialist
Varas, Oscar (RCH) Dome Mechanic
Astronomical News
A wide feld image of the southern hemisphere con-
stellations Crux and Centaurus which make up part
of the Scorpius Centaurus Association (Sco OB2),
the nearest region of recent massive star formation.
This OB association was studied by Adriaan Blaauw
in his PhD thesis (1946) and published in Bulletin
of the Astronomical Institutes of the Netherlands
(Blaauw 1952, BAN, 11, 414).
C
r
e
d
i
t
:

E
S
O
/
Y
.

B
e
l
e
t
s
k
y
51
Astronomical News
Blaauw was well known for his legendary
patience and wisdom, and for his genu-
ine interest in astronomy and astrono-
mers, including the most junior students.
He liked to bring order quietly to most
topics that he turned his attention to. This
included the archives of ESO and of the
IAU work which resulted in two books,
ESOs Early History and a History of
the IAU. His personal account of his life
entitled My Cruise Through the World
of Astronomy, published in the 2004
Annual Reviews of Astronomy and Astro-
physics, provides an extraordinarily
accurate picture of a truly remarkable
person, who infuenced the lives of many
others in a very positive way.
Tributes to Adriaan Blaauw will appear in
the next issue of The Messenger.
Based on the text of the ESO Announce-
ment ann1090.
Adriaan Blaauw, the European Southern
Observatorys second Director General,
and a key fgure in ESOs early history,
died on 1 December 2010, at the age
of 96.
Adriaan Blaauw was one of the most
infuential astronomers of the twentieth
century. I had the privilege to be amongst
his students when he returned to Leiden
from his position as Director General.
He continued to remain keenly interested
in everything to do with ESO, and still
had his characteristic twinkle in the eye
when he visited La Silla and Paranal
earlier this year. It is hard to grasp that he
is no longer with us, said ESOs cur-
rent Director General, Tim de Zeeuw.
Adriaan Blaauw was born in Amsterdam,
the Netherlands, in 1914. He studied
astronomy at Leiden University, under de
Sitter, Hertzsprung and Oort, and
obtained his doctorate with van Rhijn at
the Kapteyn Laboratory in Groningen
in 1946, with a thesis entitled, A study of
the ScorpioCentaurus cluster. During
his career, Blaauw studied the properties
of OB associations (groups of young,
hot stars) which contain the fossil record
of their star formation history. Perhaps
his most famous work explained why
some of these stars are found to be trav-
elling unusually rapidly so-called run-
away stars. Blaauw proposed that these
stars had originally been members of
binary systems, but that when one star in
the binary experiences a supernova ex -
plosion, its companion suddenly ceases
to feel the gravitational pull that keeps
it in its orbit and hence it runs away at
its orbital velocity. This work was pub-
lished in his 1961 paper, On the origin of
the O- and B-type stars with high veloci-
ties (the run-away stars), and some re -
lated problems.
In addition to his distinguished research
career, Blaauw played a central role in
setting up ESO. In 1953, the astronomers
Walter Baade and Jan Oort proposed
the idea of pooling European resources
and funding to create an astronomical
research organisation that could com-
pete on the international level. Blaauw
had returned to Leiden in 1948, moved to
Yerkes Observatory in 1953, becoming
its associate director in 1956, and moved
back to Groningen in 1957, where he was
in a key position to contribute to putting
the idea of Baade and Oort into practice.
ESO was founded in 1962, and Blaauw
took up the position of Scientifc Director
in 1968 and subsequently became the
organisations second Director General
from 1970 until 1974. During that time,
several telescopes, including the ESO
0.5-metre and 1-metre Schmidt tele-
scopes, began operating at ESOs frst
observatory site, La Silla, in Chile. This
was also a key period for the design and
construction of the ESO 3.6-metre tele-
scope, which had its frst light in 1976.
Blaauw decided that it was crucial for this
project to move ESO Headquarters
from Hamburg to Geneva, to beneft from
the presence of the CERN engineering
group. Today, the 3.6-metre telescope
remains at the forefront of research, host-
ing the HARPS spectrograph the
worlds foremost exoplanet hunter.
The early years for any organisation, par-
ticularly one in which half a dozen coun-
tries and their governments are involved,
can provide many challenges. But with
Adriaan Blaauws leadership, preceded
by Otto Heckmann, ESO started off on a
strong footing, said de Zeeuw.
After stepping down as Director General
of ESO, Blaauw returned to Leiden,
and continued to play a very important
role in international astronomy. He
was President of the International Astro-
nomical Union from 1976 to 1979, during
which period he managed to convince
China to rejoin the IAU. Blaauw retired in
1981 and moved back to Groningen,
but stayed active. He served as Chairman
of the Scientifc Evaluation Committee
for the European Space Agency satellite
Hipparcos, advising on many aspects of
its scientifc programme.
In Memoriam Adriaan Blaauw
ESO, the European Southern Observa-
tory, is the foremost intergovernmental
astronomy organisation in Europe. It
is supported by 14 countries: Austria,
Belgium, the Czech Republic, Denmark,
France, Finland, Germany, Italy, the
Netherlands, Portugal, Spain, Sweden,
Switzerland and the United Kingdom.
ESOs programme is focused on the
design, construction and operation of
powerful ground-based observing
facilities. ESO operates three observa-
tories in Chile: at La Silla, at Paranal,
site of the Very Large Telescope, and at
Llano de Chajnantor. ESO is the Euro-
pean partner in the Atacama Large Mil-
limeter/submillimeter Array (ALMA)
under construction at Chajnantor. Cur-
rently ESO is engaged in the design of
the 42-metre European Extremely Large
Telescope.
The Messenger is published, in hard-
copy and electronic form, four times a
year: in March, June, September and
December. ESO produces and distrib-
utes a wide variety of media connected
to its activities. For further information,
including postal subscription to The
Messenger, contact the ESO education
and Public Outreach Department at the
following address:
ESO Headquarters
Karl-Schwarzschild-Strae 2
85748 Garching bei Mnchen
Germany
Phone +498932006-0
information@eso.org
www.eso.org
The Messenger:
Editor: Jeremy R. Walsh
Design: Jutta Boxheimer; Layout, Type-
setting: Mafalda Martins and Jutta
Boxheimer; Graphics: Roberto Duque
www.eso.org/messenger/
Printed by Peschke Druck
Schatzbogen 35, 81805 Mnchen
Germany
Unless otherwise indicated, all images
in The Messenger are courtesy of ESO,
except authored contributions which
are courtesy of the respective authors.
ESO 2010
ISSN 0722-6691
Contents
The Organisation
G. Filippi Enabling Virtual Access to Latin-American
Southern Observatories 2
G. Witzel et al. On the Instrumental Polarisation of NAOSCONICA 5
P. Hammersley et al. Upgrading VIMOS 8
R. Arsenault et al. Progress on the VLT Adaptive Optics Facility 12
L. Testi et al. ALMA Status and Progress towards Early Science 17
A. Seifahrt et al. Precise Modelling of Telluric Features in
Astronomical Spectra 21
M. Sterzik et al. Astronomy Meets Biology:
EFOSC2 and the Chirality of Life 25
R. Gratton et al. Observations of Multiple Stellar Populations in
Globular Clusters with FLAMES at the VLT 28
S. Clark et al. Dissecting the Galactic Super Star Cluster
Westerlund 1 A Laboratory for Stellar Evolution 31
R. Maiolino et al. AMAZE and LSD: Metallicity and Dynamical Evolution
of Galaxies in the Early Universe 36
Astronomical News
J. Walsh Raymond Wilson Honoured with Two Prestigious Prizes 41
G. Lo Curto et al. Availability of Reduction Software for HARPS Data
at ESO Headquarters in Garching 42
O. Sandu, L. L. Christensen ESO Participation at the Joint European and
National Astronomy Meeting in Lisbon, Portugal 42
M. West Visiting ESOs Offce in Santiago 44
H. Boffn et al. Caf & Kosmos Events in Munich 44
New Staff at ESO J.-P. Berger 45
Fellows at ESO M. Moerchen, D. Krajnovi 47
T. de Zeeuw In Memoriam Christine Nieuwenkamp 48
Announcement of the ALMA Community Days: Towards Early Science 49
Announcement of the ESO Workshop Fornax, Virgo, Coma et al.:
Stellar Systems in Nearby High Density Environments 49
Personnel Movements 50
In Memoriam Adriaan Blaauw 51
Front cover: Colour image of the barred spiral galaxy NGC 1365 (type SBb) in the
near-infrared, taken with HAWK-I on the VLT. NGC 1365 is at a distance of about
18 megaparsecs and hosts an active galactic nucleus. The picture was created
from exposures through Y, J, H and K flters with exposure times of 4, 4, 7 and
12 minutes respectively. See Photo Release eso1038 for more details.

Messenger No142

Transféré par

Informations du document

Description originale:

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Messenger No142

Transféré par

Droits d'auteur :

Formats disponibles

The Messenger

No. 142 December 2010

Vous aimerez peut-être aussi